Discovery & People

**Interview Details** - Date: 2026-02-24 - Interviewer: Connor Pundick - Role: Inventory Control - Location: Vancouver Office - Duration: ~54 minutes - Source: Jamie transcript + Connor post-call debrief **Summary** Dermot is Bocci's Inventory Control lead — the person who manually bridges the physical world (glass production, assembly, receiving) with the digital record in NetSuite. He's tech-forward, quietly competent, and genuinely open to automation. He doesn't feel frustrated by his manual work, but he clearly wants more time for analysis and less for reactive data entry. Connor's read: I was underprepared for this one and feel I left information on the table. But the core signal is clear — the glass production → NetSuite → Felicia scheduling loop is fully manual and I can close it. Dermot would be a great co-builder and early adopter. Jamie speaker note: Dermot is labeled "Speaker 3" throughout. "Speaker 1" at a few points appears to be mis-tagged Connor turns. "Speaker -1" at 18:21 is a brief Dermot response. **Role & Responsibilities** Joined November 2022 from Janes (denim retailer). Wanted to move into analysis work. Came in as the company was transitioning from paper picking to barcode scanning — has been part of that digitization journey ever since. Core daily responsibilities: - Process assembly sheets (paper tally marks from downstairs assembly team) → NetSuite assembly builds - Receive hot shop output via Google Form → Google Sheet → NetSuite inventory adjustments - Manage Surrey↔Vancouver transfers (2x/week inbound; every 2-3 weeks outbound) - Coordinate raw glass (collet) transfers from Surrey to hot shop via email with Felicia - Bi-weekly: NetSuite export → Excel analysis → production quantity requests to Felicia Smaller responsibilities: - Vendor invoicing for received shipments (small finance slice vs. Lani's broader purchasing role) - Bin creation in NetSuite when new items need shelf locations - QC discrepancy resolution (picking errors, count mismatches) - Overstock analysis on ad hoc basis for floor team **Key Findings** **Finding 1: The Production Feedback Loop Is Fully Manual** The core inventory cycle: - Hot shop blows glass - Vivi/Vanessa QC upstairs (28s), or assembly team QCs downstairs (84s, 118s) - Paper tally sheets (A/B/garbage columns) drop into Dermot's physical inbox tray - Dermot manually enters results into NetSuite (inventory adjustments for raw glass, assembly builds for hardware absorption) - Every 2 weeks: Dermot exports NetSuite → Excel analysis → sends production quantity requests to Felicia - Felicia builds hot shop schedule from those requests; Dermot and Felicia negotiate back on priorities Connor's read: This is a complete feedback loop that is entirely manual at every handoff. The data exists — it just lives on paper and spreadsheets before Dermot re-keys it. Closing this loop (automated counts → NetSuite → Felicia dashboard) is one of the highest-ROI inventory interventions available. **Finding 2: Two Input Formats, Same Manual Bottleneck** Downstairs (assembly): paper tally sheets → physical tray → Dermot keys into NetSuite. Hot shop (upstairs): Google Form → Google Sheet → Dermot pulls and enters NetSuite. Dermot prefers the Google Form: "It shoots onto a Google Sheet straight away — easier to export or pivot table." Both still route through him. Constraint: downstairs team has no desktop access — they work with their hands. Any digitization there needs a non-desktop solution (tablet, scan gun). **Finding 3: Portland (Elements Glass) — No Feedback Loop, Easy Fix** ~3 pallets/week arriving from Elements Glass (28s, 18s, 87s, 38s). Dermot has PO visibility at a color/month level but no week-by-week schedule or specific quantities by date. When asked what one thing he'd fix: "I'd love to control what they made. If I could get fine detail — this quantity of this pendant by this date — I could be more confident in answers to sales." When Connor pushed on why he hasn't done it: "The more I talk about it, I think Tyler would just say 'Yeah, just email them.' I think I'd just do it." Connor's read: He identified the fix himself and never pulled the trigger. No tech required — just a structured weekly production report request to Elements. A template and a nudge. **Finding 4: NetSuite Partial Fulfillment Save Bug** When a large order has 99 of 100 pieces ready, the team sets the 99 aside — but can't save the partial fulfillment state in NetSuite. Those 99 allocated pieces stay visible as available inventory until the full order is scanned at fulfillment. James (sits next to Dermot, sends orders to the scanner) can inadvertently double-promise against them. Dermot's framing: rare, not a burning issue. He manually spot-checks before confirming to sales on high-value or low-quantity orders. Connor's read: Real data integrity gap — just hasn't caused a serious incident yet. Being worked on with Prolepto. Worth flagging to Andrew/Tyler as an active risk. **Finding 5: Receiving Is the Highest-ROI Barcode Opportunity** Current scanning state: - Upstairs picking: scan shelf location, input quantity pulled — no item-level barcodes on boxes - Downstairs assembly: fully manual, no scanning - Surrey receiving: fully manual — print packing lists, tick items, Dermot re-enters virtually Dermot: "On the receiving side — if there was a label you could scan, it would eliminate errors and speed things up. That's the biggest ROI." Has prior experience with full item-level barcode scanning. Knows what it looks like and wants it. **Finding 6: Collet Tracking Is Off-System** Raw glass (collet) for the hot shop is tracked on Felicia's Google Sheet — not in NetSuite. Dermot coordinates transfers by email. Functional but siloed. Another data source outside the system. **Finding 7: Lani's Return Reset Some Digitization Progress** When Dermot started, a previous person was moving the operation toward more system-based processes. When Lani returned, she reverted to trusted Google Sheets and paper trail methods — Dermot: "more clicks, but it's what's worked for her for years." Andrew called this out as Dermot = new school, Lani = old school. Dermot was diplomatic but confirmed: the reversion added complexity. Connor's read: Don't position inventory automation as "fixing Lani's process." Frame it as building on what Dermot already started with the Prolepto dashboards. **Finding 8: Dermot Wants More Analysis, Less Reactive Entry** Most satisfying work: the bi-weekly scheduling analysis with Felicia — "when it works out and there are no gaps for months, that's satisfying." He hasn't had a chance to be proactive because he's always reacting to shortages. New product launches and large regional orders (Milan, New York) keep consuming hot shop capacity, resetting his ability to build reserves. **Change Readiness: 🟢 Champion** Advocated for barcodes in his own job interview. Co-built Prolepto dashboards. Has prior WMS barcode experience. Uses ChatGPT for NetSuite reports. Explicit: "I'm all for it — just something that makes my job easier." Approach: treat as co-builder. Bring him into design conversations early. He will be an internal advocate. **Opportunities Identified** **Action Items** **Connections** - Felicia VanDenBoogaard: Direct collaborator on bi-weekly hot shop scheduling. Same capacity constraint picture from her side. - Andrew Nguyen: Endorsed Dermot as tech-forward. Aware of new-school/old-school dynamic. Wants barcode expansion. - Lani Anthony: Interfaces on inventory reconciliation. Different change readiness profile. Reverted some digitization work. - James (fulfillment): Sits next to Dermot; most affected by the partial fulfillment NetSuite bug. - Conor Graham: NetSuite system owner — any inventory system changes flow through him.

**Interview: Bronwyn Livingston — Office Manager / Travel & Culture** Date: 2026-02-23 Duration: ~60 min Source: Jamie transcript (2026-02-23) + Connor Pundick post-call debrief **Pre-Interview Context** Role: Office Manager and Director of Travel & Culture. Manages customer database and email lists using Klaviyo — formats newsletters, schedules campaigns. Books all company travel (~100 flights/year), handles event planning, manages Galiano Island cabin perks, and serves as office/facility manager. Long-tenured (16 years) multi-domain generalist. Reports to: Lana Dooling (General Manager) What we knew going in: - Lana described her as "operational marketing support + office manager" — Klaviyo, newsletters, travel, events, supplies - Sole Klaviyo owner across the company - The change management gap has a direct line to Bronwyn: if product changes don't reach her, they don't reach the email list - Johannes (Berlin) leads communications direction — how much day-to-day direction Bronwyn receives was an open question - Daria leads campaigns with external partners — coordination with Bronwyn's email work was unclear Connor's read going in: One of the handful of people at Bocci who manages broadly rather than in a lane. Expected institutional memory, systems in her head. More dynamic interview than most. **Role & Responsibilities** Bronwyn joined 16 years ago from a Craigslist posting when Bocci had 13 people. Worked under Anu for 4 years before Lana took over; credits the culture shift to Lana. Role has expanded organically from pure admin to a wide multi-domain function: - Travel management: ~100 flights/year. 37 booked in one recent month during Milan planning. Covers Jay and Tyler (Czech Republic ~monthly), Glassblop team (Portland), Jaden (door fabrication), Paul (installations), full sales team for Milan. - Event management: Annual 20th anniversary party, October warehouse sale, IDS/Design Festival Vancouver, Milan Design Week/Salone. Most satisfying part of her role. - Klaviyo / email marketing: 30,000 contacts. Manages database, segmentation, scheduling. Content from Darya/Mandy; VIP lists from Erin (PR). - Office/facility management: Building manager. Currently planning a 6-10 month renovation starting within ~12 weeks. - Galiano cabin perks: Manages two company-owned cabins on Galiano Island — calendaring, cleaning schedules, caretaker coordination, key/access management. - General admin: Supplies, Expensify reconciliation, phone plan updates. **Key Findings** **1. Travel Management Is the #1 Pain Point** 100 flights/year, all managed manually. Core constraint: Randy's Air Canada points sit in a personal account — only Bronwyn can access it. She tried delegating to a travel agency; the agency couldn't adapt to Bocci's complexity, she took the work back. Booking spans 6 platforms with no integration: Air Canada, BA, KLM, hotels.com, Airbnb, Booking.com. Tracking via email, manual screenshots, and a spreadsheet that sometimes lags. Expensify adds friction: two company cards with different forwarding addresses, currency conversions, tips creating separate receipts. At time of interview: Tyler and Andrew's flights canceled due to a New York storm ahead of the 93 launch event Thursday. Emergency rebooking mid-conversation. Connor's read: She mentioned travel management tools multiple times unprompted — clearest ROI, most immediate enthusiasm. She's explicitly asking for help. Action item: evaluate tools that unify flights + Airbnb + Booking.com + points optimization. **2. Institutional Knowledge Lives in Her Head — By Necessity** Bronwyn has built her entire operating system around Randy and Omer's personal preferences. This isn't a failure of documentation — it's structural. Randy likes live music, indoor/outdoor vibe, hot shop for food, multiple bars, good wine, no drink lines. She knows which vendors barter: Shambar, Tando, Bodega Ridge, Harvest, Small Victory. She knows Randy calls her daily. None of this can be abstracted into a workflow. Connor's read: Same dynamic as Tyler — a person absorbing complexity the organization can't systematize. The difference: Tyler's knowledge is about information routing; Bronwyn's is about preferences and relationships. The cabins, vendor barters, event muscle memory — all in one person after 16 years. Both are single points of failure. **3. Galiano Cabin Perk — A Hidden Admin System** Bocci owns two cabins on Galiano Island. After 2-3 years of tenure, employees get 1 weekend/year; after 5 years, 2 weekends. Bronwyn manages all of it: booking calendar, cleaning schedule, caretaker coordination, key/access management. Not mentioned in any prior interview. Classic "one more thing" that turns out to be a complete system running solo. **4. Klaviyo — Well-Run, Event-Driven, Limited by Content Inputs** Moved from Mailchimp this year — significantly better results. Lower unsubscribe rates, higher engagement. Tracks direct sales impact (30% CTR → $10K in product sales in one campaign example). Cadence is event-driven: launches, shows, parties. Four upcoming parties plus the 93 New York launch. Segmentation uses Erin's (PR) VIP lists — mix of celebrities, architects, dealers, younger demographics. RSVP tracking via Typeform → Klaviyo separate list. Content/design split: Bronwyn owns contacts/database/scheduling; Darya and Mandy handle design. Information flow gap to flag for Erin interview: Bronwyn's email content depends on timely product information from Erin and the design team. Given the company-wide change management gap (no product change ledger), it's likely that email content sometimes lags reality. Not confirmed directly in this interview — probe with Erin. **5. System Adoption Pattern — She Sees the Risk Clearly** Previous implementations failed due to adoption, not system quality: - Slack: not used to full capacity - Swivel: image/video database — adoption failure - Travel agency: couldn't adapt to Bocci's complexity, she took the work back Her framing: "If old systems aren't completely shut down, people stay on them." Any new tool needs to be easy, well-thought-out, and trainable. Connor's read: She's a champion because she understands the failure pattern and is still asking for help. That's more useful buy-in than uncritical enthusiasm. **6. Upcoming Renovation — 6-10 Months on Her Plate** Major renovation starting within ~12 weeks: hot shop expanding, cold shop moving to basement, kitchen relocating, all windows/doors replaced, electrical transformers moving outside (frees up small meeting room). Bronwyn is managing this as building manager on top of everything else. Will reduce capacity for new system onboarding during this window. **7. Team Context** - Mandy: bottleneck confirmed — many people competing for her time, room to hire another designer - Ally: took on PM + CAD/LATAM sales when Anna went on mat leave; currently in Toronto (father had heart attack), stretched thin - Erin (PR): too busy, but Connor has 30 min Friday or Monday — she manages VIP lists and product launch info flow **Post-Interview Capture** Top 3 pain points: - Travel management — 100 flights/year, 6 disconnected platforms, no delegation possible due to Randy's personal accounts - Scattered focus / context switching — constant interruptions (facility issues, emotional support, admin requests), focused work near-impossible - Email content pipeline — dependent on timely product info from Erin/design team; likely lags reality given change management gaps Actual tools used: Klaviyo, Typeform, Expensify, Air Canada (personal login), BA (Amex account), hotels.com, Airbnb, Booking.com, Google Sheets, Slack, email Change readiness: 🟢 Champion — open to change, realistic about adoption risk, explicitly asking for help on travel Quick-win candidates: - Travel management portal unifying flights + Airbnb + Booking.com + points - Galiano cabin calendar/booking system (simple shared calendar with booking rules) - Email content pipeline from ProDev/PR → Klaviyo (feeds off the change ledger once live) Surprises / unexpected insights: - Galiano cabin perk system — entirely managed solo, not mentioned in any prior interview - Travel volume (100+ flights/year, 37 in one month) — higher than expected - Renovation scope: 6-10 months starting in ~12 weeks — will affect her availability **Action Items** Source transcript: Bronwyn Interview Jamie note: "Speaker 1" throughout = Bronwyn Livingston.

**Interview Details** **Key Findings** **1. Glass Shop Operates as an Isolated System** Felicia has no access to NetSuite. She has never used it. The entire glass shop — scheduling, inventory, purchasing, vendor management — runs on Google Sheets she built herself while teaching herself the tool. Nothing flows into or out of the company's ERP. Connor's read: This is a black box. The glass shop's inputs, outputs, and tracking are completely invisible to the rest of the organization. There is no real-time visibility into production status, material consumption, or inventory levels for anyone outside this function. Integrating this into NetSuite or a shared data layer is a high-value opportunity — signal intelligence from a database would eliminate significant manual work. **2. Color Bar Inventory Tracking — Complex Manual System** Felicia tracks 30+ colors of color bar across three vendors: - Kreugler (via Hot Glass Color) - Gaffer / Reichenbach (via Olympic Glass Rods) - Bullseye (direct) For every color, she tracks: - Manufacturer and batch number - Usage rate per pendant (bodies vs. satellites — different consumption amounts) - Current inventory with 6-month (orange) and 12-month (red) thresholds - Batch quality testing results against golden samples Complication: Post-COVID and the Ukraine war, color bar suppliers shifted mineral sources, causing tone inconsistencies. Felicia now tests every incoming batch against golden samples. Some colors (e.g., K53/gray two) have a history of recurring tone shifts. When a color goes off-tone, they may source from alternate manufacturers — which means tracking two bars for the same color with different density/usage rates. Felicia's concern: Some calculations in her spreadsheet may not be accurate. She taught herself Google Sheets and suspects some formulas have errors. She specifically asked Connor to review them. **3. Glass Consumption Tracking — Compliance Gap** Glass usage (cullet) is tracked on a weekly schedule spreadsheet. Glassblowers are supposed to log furnace charges on a shared computer in the hot shop, but people often forget. This means consumption data is unreliable. **4. Production Scheduling** Felicia creates a weekly production schedule in Google Sheets. Takes 30 minutes to 2 hours depending on: - Complexity of the week's production mix - Interruptions and last-minute changes - Whether she gets a solid block of uninterrupted time Schedule inputs come from Dermot (2-week production requests) and James Gibbs (order fulfillment priorities). Felicia negotiates back: "We can do this, but we can't do that." **5. Elements — External Manufacturing (Portland)** Bocci outsources 28 production to Elements, a glass studio in Portland. Felicia manages this relationship entirely through a shared Google Sheet and email. She has never met anyone from Elements in person. Key contacts: Aaron Frankel (owner), Ella (operations), Ian (owner, less frequent contact) What Felicia manages: - Shipping or ordering color bar for Elements - Tracking their color bar inventory (target: 7-15 days of stock) - Monitoring PO fulfillment — what's been made, what's on palettes, what's shipped - Coordinating Oceanside cullet orders through Conor Graham when Elements runs low Biggest pain point (self-identified): The Elements tracking requires flipping between multiple tabs — color inventory, outstanding PO, palette tracking — to assemble a complete picture. She wants a rollup view that consolidates this. Elements manages their own Bullseye cullet orders directly with Bullseye, but relies on Felicia/Bocci for Oceanside cullet. **6. Hot Shop Infrastructure** Key constraint: A furnace takes ~1 week to heat from cold (limited by crucible — ceramic bowl that cracks if heated too fast). This is why furnaces run 24/7, including weekends. **7. Team & Staffing** - Glass shop team: ~9 people (excluding Felicia), mix of full-time and part-time - Hot shop: 2-6 people on a given day - Cold shop: 1-3 people - Weekend workers: Steve (Saturday, cold work + furnace charging), one person Sunday - Equipment maintenance: Jesse (primary), Felicia (backup/emergency) **8. Material Supply Chain** - Oceanside cullet: Ordered ~twice/year in 40-foot containers. Conor Graham handles ordering. Stored in Surrey warehouse, one palette kept in-house behind Andrew's desk. - Bullseye cullet: Felicia orders directly from Bullseye, Conor gets the invoice. - Color bar: Felicia orders directly from vendors. Credit card on file, invoices go to Conor Graham. - Elements (Oceanside): Felicia coordinates through Conor when Elements runs low. **Key Interfaces** **Tools Used** - Google Sheets — scheduling, color bar inventory, glass consumption, Elements tracking (all self-built) - Email — vendor communication, Elements coordination - Slack — on phone only, personal email account (not Bocci email). Needs account migration help. - No NetSuite access - No other enterprise tools **Action Items** **Follow-Up Topics (for next session)** - Complete the remaining interview questions (process mapping, additional pain points) - Deep-dive on Elements workflow — map the full cycle from PO to delivery - Understand the daily tracking sheet in more detail - Explore scheduling optimization opportunities - Map the full purchasing cycle end-to-end - Understand prototyping documentation workflow **Change Readiness Assessment** Connor's read: Felicia is a strong champion for change. She actively asked for help with her processes, acknowledged her spreadsheet limitations, and was enthusiastic about the engagement. She's not protective of her current tools — she knows they're held together with duct tape. Her 8 years at Bocci and deep domain knowledge of glass production make her an invaluable partner for any system improvements in this area. Assessment: 🟢 Champion

**Jennine Banks — Accounts Payable** Date: Feb 18, 2026 | Duration: ~30 min Reports to: Lana Dooling | Location: Vancouver Office Interviewer: Connor Pundick Context: Relatively new. Per Conor Graham: handles front-end/busy work including repetitive tasks like payment matching that should be automatable. Conor recommended sitting with Jennine for an end-to-end sale process walkthrough. **1. Role & Context** - Typical week: Morning: payment processing and email review across 4 bank accounts (Deutsche Bank + 3 CIBC). Afternoon: vendor invoice entry into NetSuite. Ongoing: FedEx customs disputes. - How long at Bocci: Relatively new - Works most closely with: Lana (approves vendor invoices, processes payments Thursdays with Conor Graham, handles FedEx customs disputes), Conor Graham (approves sales orders, determines regional shipping charges), Tyler (approves vendor invoices), Michelle (provides commercial invoices for customs disputes), Megan (shipping, currently on vacation) - What they enjoy most: Not explicitly stated — she sees the work as "the job" - Org structure: No single direct manager — collective leadership between Randy, Lana. Team operates as self-starters. Connor's read: Jennine doesn't realize how much of her work is automatable — she thinks it's just the job. Connor estimates 60-70% of her workflows could be automated. The sheer volume of manual work she's doing was shocking. Notes: Jamie labeled Jennine as "Speaker 0" throughout the transcript. **2. End-to-End Sale Process** **Cash Application & Payment Processing (~50% of her day)** - Processes customer payments daily from four bank accounts: - Manual reconciliation: eyeball-matches NetSuite balances against actual bank balances - Matches payments to invoices using proforma details — clients instructed to include invoice number with wire payments, but compliance varies - If info missing: searches by client name or amount, contacts sales team to narrow down - Payment match rate is high — only a few discrepancies per month - ProForma invoices list currency-specific banking info at bottom (Euro → Deutsche Bank only, CAD/USD → CIBC), occasionally causing currency mismatches **Vendor Invoice Processing (afternoon workflow)** - Enters vendor invoices into NetSuite, uploads invoice documents to vendor profiles - Lana and Tyler review/approve on their own schedule - Lana and Conor Graham execute payments on Thursdays - Once approved, Jennine processes and releases payments **Shopify & Stripe Processing** - Shopify: Auto-uploads to NetSuite with merchant fees typically correct. Reviews for refunds and canceled orders. Refunds don't flow back automatically from NetSuite to Shopify. - Stripe: Entered manually by Sachie and Emily, approved by Conor Graham. Jennine creates deposits when payment appears in bank. Occasionally orders missed by entry team. No NetSuite integration — unlike Shopify. Connor's read: This is the automation goldmine. The cash application workflow alone — downloading CSVs, eyeball-matching across 4 accounts, manually reconciling — is textbook automation. The Stripe gap is a known issue that mirrors what Emily and Sachie described from the sales side. **3. Communication & Decisions** - Goes to sales team when payment info is missing from the system - Generally has the information she needs — describes core processes as "90% information availability" - Pain points concentrated in FedEx customs (external constraint) and expense reporting discipline (people problem) **4. Tools & Systems** - Tools used daily: NetSuite (central accounting), Gmail, Excel (bank CSV downloads), Deutsche Bank portal, CIBC online banking, Stripe, Shopify - Tools supposed to use but don't: None mentioned - Workarounds built yourself: Manual eyeball reconciliation across bank statements and NetSuite - Magic wand fix: Not explicitly stated - AI tool usage: None mentioned **5. FedEx Customs & Duties Pain Point (Major)** - Handles DDP shipments where Bocci pays duties and customs - CUSMA/USMCA covered shipments should not incur customs charges but FedEx systematically miscodes tariffs - Volume exploded: from 1 customs invoice/week to 5-6/week - Estimates ~2 hours/week on customs disputes alone - Must dispute nearly all invoices — usually wins, but FedEx often requests more information - $1M backlog in suspense account (2023-2024 disputes) awaiting FedEx resolution — some years old, FedEx slowly issuing credits - FedEx transportation invoices: 400-500 shipments/week requiring manual review **6. Expense Report Management** - Uses Expensify for employee expense reporting - Chronic delinquency from same employees — missing itemized receipts, screenshots showing only payment confirmation, missing GST info - Delays GST submissions and creates audit risk - Exception: Jay takes expense reporting seriously **Post-Interview Capture** - What did Jennine care about most? Getting through her daily workload. She doesn't frame it as pain — it's just what she does. - Top 3 pain points: - Actual tools used: NetSuite, Gmail, Excel, Deutsche Bank portal, CIBC online banking, Shopify, Stripe, Expensify - Change readiness: Open — she's not resistant, but she doesn't see the automation opportunity. She thinks the core processes are already "relatively seamless at 90%." - Quick wins: - Surprises: The FedEx $1M backlog is significant. Also the sheer manual nature of the cash application — this is 2026 and she's eyeball-matching CSVs. - Confirms or contradicts prior interviews: - Key quotes: - New stakeholders identified: Source transcript: Jennine Banks Interview (Jamie) — Note: Jamie labeled Jennine as "Speaker 0" Action item: Connor needs NetSuite access to shadow Jennine and document the manual processes end-to-end — this is prerequisite to building the automation case.

**Interview: Aaron Miller — Product Dev (22 Systems / 3D Modeling)** Date: 2026-02-23 | Duration: ~90 min Reports to: Brian Magennis | Location: Vancouver Office Source: Jamie transcript (2026-02-23) + Connor Pundick post-call debrief **Pre-Interview Context** Role: Industrial designer specializing in mechanical design, hardware/suspension systems, optical and electrical engineering. 7 years at Bocci. Works across both product lines (22 Systems + Bocci Luminaire). Manages global contractors (Costa Rica, India) for firmware and embedded development. Currently juggling 6 major projects. Reports to: Brian Magennis What we knew going in: - Brian recommended Aaron specifically for technical/AI perspective — strongest technical user on ProDev - Jasmine credits Aaron with introducing ChatGPT to the team - One of only 2-3 of 7 who actively use 3D Experience — his take was considered the tiebreaker - €60K sunk in 3DX migration. Brian leaning toward PDM Standard/Pro as replacement. - Jasmine warned to expect 1.5-2 hours — strong opinions, goes deep Connor's read going in: May be the smartest person on the ProDev team. Unique in that he goes deep on actual engineering — not just design thinking. Expected a technical, wide-ranging conversation. **Key Findings** **1. 3D Experience Is Dead — and Daniel Has a Conflict of Interest** Aaron considers 3DX one of the worst programs he's ever used. Save times run 30 seconds to 5 minutes per file. It crashes SolidWorks regularly even when not actively in use. The UI hides what it does under the hood — when something breaks, it's hard to trace. It requires more diligence from users, which means more mistakes, not fewer. Aaron, Jasmine, Vince, and Patty turn it off whenever possible. Brian is forced to use it. Julien uses it. The replacement is obvious to Aaron: SOLIDWORKS PDM. It's simpler, more functional, battle-tested — every engineer using SolidWorks at a design level uses PDM. The local server is already set up and ready to go. He's never heard of anyone switching to 3DX from PDM because they prefer it. The Daniel problem: Daniel is pushing hard against switching to PDM and has never given a good technical reason. Aaron's read: Daniel works for SolidWorks/Dassault, recently gave a presentation at a Dassault conference, and Bocci is one of the larger pilot companies on the cloud version of 3DX. Dassault has invested in Bocci as a reference customer — Daniel gets on calls with them a couple times a week because they want his input on development. Aaron suspects bias but doesn't want to formally accuse. Connor's read: This reframes Daniel entirely. His advocacy for 3DX isn't a product opinion — it's structurally compromised. He's the only voice defending a €60K tool that nearly the entire team has rejected, and he has a direct professional relationship with the vendor. The decision on PDM vs. 3DX shouldn't require Daniel's sign-off. **2. No Product Version Tracking — Aaron Built His Own System** NetSuite and all internal systems have no way to track which product version was shipped to which customer. When 14P units come back with water damage failures, Aaron physically inspects the processor to determine the version. His workaround: manufacturers put different colored dots per production run. Aaron maintains a personal sheet mapping dot color → processor type → firmware version → PO number. That's the only version history that exists. This means there's no data-driven feedback loop. Design changes can't be correlated to field failures without Aaron manually doing it. No one else could do this without him. Connor's read: This is exactly the kind of institutional memory risk that runs through the whole company. Aaron has systematized it for himself, but it's invisible to everyone else and dies if he leaves. **3. Gate Meetings Abandoned — Aaron Has Done This Before** Historically Bocci held formal 2-3 hour weekly product reviews with shipping, inventory, GM, and leadership covering all ~20 active projects. These became unwieldy with too many projects per meeting and were quietly dropped. Now it's ad-hoc — "Hey Omair, are you around?" Aaron has implemented formal gate meetings at two previous companies (including Regency Fireplaces). His model: a master checklist of everything that could come up in product development. Each department marks items Done, N/A, or Pending at defined checkpoints. In the fireplace industry it was ~20 items per department. This is how you catch missing packaging, missing packing lists, wrong screw hardware — currently discovered at production rather than at review. Multiple people have pushed for this at Bocci. Resistance manifests as changing the subject, not talking about it, or actively walking away. No one wants to be the bad guy, so problems get ignored instead of addressed. Connor's read: Aaron is the first person from the ProDev side to express genuine frustration with the lack of downstream accountability after handoffs. Everyone in support, sales, and documentation has described the handoff gap from the receiving end. Aaron is describing it from the sending end — he hands something off and has no visibility or accountability mechanism for what happens next. That's a strong, bilateral case for stage gates. **4. Founder Absolute Authority — Creative Freedom and Wasted Months** Bocci's founders (Randy and Omer) have absolute creative control. Either can terminate an idea or spin one up at any time, with no formal review process. Randy emails Aaron every couple of weekends and "low priority" prototypes become immediate priorities. Aaron has spent months on efforts that were entirely reversed by a single founder decision. Aaron holds both feelings simultaneously: frustration at the oscillation and wasted time, and genuine appreciation for the creative freedom it affords him. He stays at Bocci primarily for the flexibility and work-life balance — the same openness that costs months of effort is also what lets him explore, invent, and try novel things. He's the one building patent-pending capacitive touch technology and training neural networks inside light fixtures. The deeper issue: when founder decisions create direction changes, there's no accountability mechanism downstream. Other departments (sales, support, Mandy) absorb those changes without formal notification. Connor's read: This is the same "creative engine" dynamic documented throughout the engagement — but hearing it from inside ProDev adds texture. It's not just that information doesn't flow downstream. It's that the source of change (founder whims) is itself unpredictable and untracked, which makes any downstream system inherently reactive. **5. Anecdote-Driven Crisis Response — No BI, No Failure Rate Data** A single customer complaint, raised in front of the wrong people, can trigger a 3-month company pivot. Aaron described Randy calling him at 7pm about an urgent issue — Brian then spent a month producing a fix — before the sales team revealed there had been only one complaint. The company has no mechanism to contextualize individual failures against volume. Aaron and Andrew (production) both independently flag this. You can't evaluate whether a company reaction is proportionate without knowing: how many units shipped, how many came back, what version, what changed between runs. None of this is tracked. **6. AI Depth — the Most Technically Advanced User in the Company** Aaron uses AI at a level no one else at Bocci approaches. He trained a neural network that lives inside the 14P, managing complex functionality. He uses Midjourney and Stable Diffusion with ControlNet for rapid design exploration — generating 500+ concept images per session to eliminate dead-end directions faster than physical prototyping allows. He understands how LLMs work well enough to know when a model is wrong and how to bias its inputs. His company-wide LLM recommendation: - Claude (primary) — leads on benchmarks, B2B revenue 6-7x over ChatGPT, persistent memory, Claude Desktop file drag-and-drop makes it accessible to non-technical users - Gemini (strong secondary) — best Google Cloud integration, strongest on image tasks, free - ChatGPT (acceptable but not preferred) — most user attachment, but the new model's sycophantic behavior frustrates him For image generation at company level: Midjourney is the only realistic option to hand off. Stable Diffusion is too complex and specialized — he would never pass it to anyone else here. Connor's read: Aaron is the person to involve in any AI rollout design. He's not just an enthusiast — he has a working mental model of how these systems work and where they break. He also asked to sit with the open question of how ProDev should redesign file management, change communication, and AI integration if given full authority. That conversation is worth having. **7. Cross-Department Communication — Structural, Not Personal** The design team is in the back of the building. Sales is in front. Production is downstairs. Bocci used to do desk swaps every 3 months — rotating seating to force inter-department contact. They had to stop as the company grew. Aaron notes that inter-department communication has been a big issue at every company he's worked at. At Bocci, production downstairs is especially isolated: "it's really up to us to go seek them out." He walks the production floor periodically and finds things like: a screw that's been painful to install for 8 years that he could fix by making it 2mm shorter — but no one told ProDev. The information flows when he physically walks down, not through any system. **Post-Interview Capture** What Aaron cared about most: Getting off 3DX and onto PDM. The accountability gap at product handoffs. Founder-driven direction changes wasting ProDev effort. Top 3 pain points: - 3D Experience — productivity drain, daily friction, team has largely abandoned it, blocked from switching by Daniel - No version tracking or failure correlation — homemade colored-dot system is the only version history that exists - Lack of formal gates / handoff accountability — changes made ad-hoc, downstream teams not notified, no one wants to be the bad guy Actual tools used: SolidWorks, 3DX (reluctantly), Windows Explorer (primary file management workaround), Claude, Midjourney, Stable Diffusion + ControlNet, ChatGPT, Gemini, personal spreadsheet for version tracking Change readiness: 🟢 Champion — technically sophisticated, explicitly asks for better systems, has implemented gate meetings before, willing to engage further Quick-win candidates: - Migrate from 3DX to SOLIDWORKS PDM (server already set up — this is unblocked except for Daniel) - Simple version tracking field in NetSuite — PO to product version mapping - Gate meeting checklist — Aaron has a template model from previous companies Surprises / unexpected insights: - Daniel's conflict of interest — SolidWorks employee, presenting at Dassault conferences, Bocci is a reference customer. Not a product opinion — a structural bias. - Aaron is the first ProDev person to express frustration with downstream accountability. The handoff gap is bilateral, not just a downstream complaint. - Neural network inside the 14P — the technical depth here is genuinely unusual for a company of this size. - Colored-dot version tracking system — entirely self-built, invisible to the organization Confirms prior interviews: - 3DX broadly rejected (Jasmine, now Aaron — the supposed power user) - Founder bypass pattern (Brian, Tyler — now confirmed from inside ProDev) - Gate meetings needed (Brian, Tyler, Mandy — now Aaron from the ProDev sending side) - Anecdote-driven decisions without data (Andrew — now Aaron independently) - Inter-department communication as persistent structural gap (every interview) New flags: - Daniel/3DX conflict of interest — needs to be surfaced carefully in recommendations. Frame around tool performance, not accusation. - PDM migration is effectively ready — server set up, team aligned, just needs a decision - Aaron as a potential AI integration lead — strongest technical user, already thinking about company-wide deployment Open loop (sent via Slack): If Aaron had authority to redesign how ProDev manages files, communicates changes, and uses AI — what would he build? Source transcript: Aaron Interview Jamie note: Attendees confirmed as Connor Pundick and Aaron Miller. Jamie summary is accurate — Aaron speaks fast and goes on extended tangents; Jamie captured themes well despite non-linear conversation.

**Brian Magennis — Product Development Manager** Date: Feb 12, 2026 | Duration: 45 min **1. Role & Context** - Typical week: Monday starts with catching up on China team emails from Sunday night. 9:30 AM catch-up with Tyler to review projects. Tuesday morning: Daniel and Julien meeting on 22 System strategy; afternoon: Bacci team meeting. Wednesday: broader 22 System meeting with Paddy and Aaron. Thursday-Friday: open project work, though Friday is when emergencies surface. - How long at Bocci: 2+ years. Previously senior designer at Glasgow product development firm and Head of R&D at an education company. - Works most closely with: 7-person product development team — Jasmine (dedicated Bacci), Paddy (mostly Bacci), Tom, Daniel (22 System design lead), Vince (half 22 System, half Bacci hardware), Aaron (both products), and Julien (compliance contractor, 22 System product lead). Also works closely with Tyler (operations, central information hub) and coordinates with China manufacturing team (Ravi, Thomas, Rex, Lily). - What they enjoy most: The creative-to-production journey. P1 to P2 is where "a lot of magic happens." Prefers letting the China team "cook" and sending files rather than forcing integrated systems. **2. Communication & Decisions** - How do product decisions travel to the people who need them? - How do you find out about decisions Randy or Omair make? - Walk me through a recent example where a decision didn't reach everyone in time. - When you need sign-off from multiple people, how does that work today? - Do you have the info you need, or are you chasing things down? Notes: Communication breakdown is the root cause of most issues. Critical information travels word-of-mouth and can take weeks to reach relevant teams. Tyler is the central information hub, but information gets lost or misinterpreted passing through Tyler to Brian to the team — a game-of-telephone dynamic. 93 product example: The 93 product (11.5 inches, smaller version at 8 inches) created spacing challenges on certain canopy layouts because the 8-inch sizing decision was made without flagging the canopy compatibility impact. Required a special canopy series workaround. Classic case of a designer going directly to Randy without looping in the PM, so downstream consequences weren't caught. Teams make changes without informing others, later discovering different production processes have been applied. Some one-off cases trigger full redesigns while recurring issues (1 in 100) don't get addressed if manageable. Randy provides valuable feedback through friends who own Bocci products — often first to flag issues. Randy's triggers: passionate about quick idea implementation, or hearing something went wrong. No centralized tracking or formal sign-off process exists. Product knowledge is scattered across Tyler, Paddy, Paul (installers) with no comprehensive source. **3. Tools & Systems** - Tools used daily: - Tools supposed to use but don't: - Workarounds built yourself: - Magic wand fix: - AI tool usage: Notes: Tools used daily: SolidWorks (core design platform), email, Slack, word-of-mouth. 3D Experience PDM — the big one. SolidWorks-based cloud PDM platform for revision history tracking. €60K spent on migration alone (excluding software costs). Major usability problems: slow loading, poor UX, long downloads, changed drawing names requiring exact product knowledge to find files. Different users have vastly different experiences — sometimes completely non-functional. Continuous updates change functionality, creating a barrier to flow state. Daniel champions the system for revision control but hasn't brought the team along. Training became defunct after continuous product changes. Only 2–3 of 7 team members actively use it; several have logged off and use SolidWorks locally. Julian and Julien want to move away due to slowness. Workarounds: Designers work offline in SolidWorks, save locally, then push to cloud via 3D Experience (Git-like functionality). Some designers resist 3D Experience entirely, preferring local storage. Most reliable source for current files: email to manufacturing with latest drawing files, though chains continue across threads. Files live on local devices, Teams, and email. Magic wand fix: Brian recommends exploring PDM Standard or PDM Pro as alternatives to 3D Experience, but NOT changing core SolidWorks platform. Mentioned connecting with Form 3 company nearby to learn about their PDM system. Does not recommend using 3D Experience as a quick win — it's a tough problem requiring significant effort. AI tools: Half the team uses ChatGPT with personal accounts for brainstorming, design solutions, and risk analysis. ChatGPT flags specific risks the team might miss. Brian excited about introducing more AI capabilities. Senior team likely to embrace; some resistance from lighting team designers and artists. **4. Workflows & Pain Points** - Walk me through a new product from idea to production — what are the stages? - Biggest bottleneck? Give me a specific recent example. - When an existing product gets a part change, how does that flow? - Where does info get lost between your team and other departments? - Have there been attempts to formalize workflow tracking? What happened? - Hardest thing to teach a new hire: Notes: Full 18-month product lifecycle: - P1: 3D print, in-house, 2 weeks - P2: Machined sample, 2 weeks design + 3 weeks from China, 10 units for testing. P1 to P2 is where "a lot of magic happens." - T1: First tooled sample, 6 weeks, 25 units for first installs in New York and Milan - T2: Second tooled sample with finish options and installation testing by Paul, 200 units - T3: Final production version with all finishing, 300 units to showrooms and key clients before full production - Products available to customers 2 months after April release Key gap: T2 to T3 handoff. This is when wider teams (support, documentation, UL compliance via Julien) should integrate — but currently they're brought in too late. Brian recommends formalizing a handoff checklist and earlier compliance involvement to avoid costly label re-runs (€20K+). UL compliance timing problem: UL standards change every 2–4 years. Information should flow at P2 stage rather than waiting until T2. Updating labels for UL compliance can cost €20K for manufactured products. Company sometimes uses old labels then switches on next production run. Tension between adhering to updated specs and reluctance to revisit completed designs. 14P scale example: 14P has high volume (400 units monthly, targeting 1,000), making defect thresholds different than low-volume products. Recent unusual failure: aircraft anchor split on pendant, first occurrence in 12+ months of installations. Required immediate call to China for steel versions. Ideation process: Jay and Omer research global glass innovations and propose directions, with Omer as creative director narrowing multiple designer approaches to one concept. Process prioritizes reaching Omer's vision over speed, sometimes requiring complete redesigns after impasse. Brian says do NOT change this convoluted ideation-to-production workflow — it yields interesting solutions. Formalization attempts: Brian attempting to formalize spec handoff with an incomplete checklist. 22 System has better formalization: at T2 phase, a formal release happens after announcing tooling is done. Bacci side is less structured. **5. Content & Digital Assets** - What does your team provide to Mandy for spec sheets? - How does the handoff of specs/drawings work between pro dev and documentation? - Is there a single trusted source for current product info? Notes: No formal process for passing design information to Mandy for specifications. Brian is attempting to formalize with an incomplete checklist. Sometimes the team sends 3D files directly to Mandy for extracting views for specs and instructions. Mandy (industrial designer) has access to 3D Experience and navigates well enough for her needs. 22 System has better formalization: at T2 phase, a formal release happens after announcing tooling is done. Bacci side has no equivalent. No formal single source of truth across the organization. Most reliable source is email to manufacturing with the latest drawing files, though chains continue across threads. Design changes are made constantly without documentation. **6. Aspirations & Closing** - What should NOT change about how your team works? - If you were running this engagement, what would you tackle first? - Anything I haven't asked that I should know? - Anything you're nervous about? Notes: What should NOT change: The convoluted ideation-to-production workflow — it yields interesting solutions. Do not change from SolidWorks as core development system. If running this engagement, tackle first: Start with P2 to production (T2, T3) phase and loop in all necessary parties. Formalize the handoff checklist. Focus on knowledge dissemination between the whole team including sales. Formalize processes after fully understanding the physical workflow, then digitize. Other recommendations: Review and standardize AI systems across the organization. Explore PDM Standard or PDM Pro as 3D Experience alternatives. Connect with Form 3 company nearby to learn about their PDM system. Key timing constraint: Brian traveling March 16 for a full month. All engagement work needs to be concentrated before then. Next interviews recommended: Julien (UL standards/information flow), Daniel (3D Experience/design systems), Paddy (institutional knowledge), Aaron (technical capabilities/AI usage), potentially Mandy (documentation and specs handoff). **Post-Interview Capture** - What did Brian care about most? Formalizing the handoff process from P2 through T3. He sees this as the #1 lever — loop in support, documentation, and compliance earlier, give them a checklist, and you prevent the costly downstream scrambles. - Top 3 pain points: - Actual tools used: SolidWorks, 3D Experience (partially), email, Slack, word-of-mouth, ChatGPT (personal accounts) - Change readiness: Champion — Brian is actively trying to formalize processes, excited about AI, recommended specific alternatives (PDM Standard/Pro), and wants to digitize after understanding physical workflows - Quick wins: T2–T3 handoff checklist with sign-off gates; AI tool standardization across the team (half already using ChatGPT); earlier compliance gating at P2 stage instead of T2 - Surprises: March 16 travel deadline — Brian gone for a full month, so all engagement work must be concentrated before then. Also: the 93 product canopy issue revealed a designer→Randy bypass that completely skips the PM, meaning downstream consequences weren't caught. - Confirms or contradicts Molly/Tyler: Confirms Tyler's assessment of communication fragmentation — Brian independently identifies Tyler as central hub where information gets lost or misinterpreted. Confirms the change management gap Lana identified as the #1 company-wide need. Adds specificity: the gap is sharpest at T2–T3 where wider teams should integrate but don't. - Key quotes: Source transcript: Brian Magennis Interview: Product Development Workflow

**Conor Graham — Finance / NetSuite Admin** Date: Feb 12, 2026 | Duration: 45 min **1. Role & Context (5 min)** - How long at Bocci: - Typical week: - Works most closely with: - What they enjoy most: Notes: - 14 years tenure as of February 20 — longest-tenured systems person at Bocci - Fills three distinct roles accumulated over time: de facto head of finance, shipping department oversight, and systems architecture. "Kitchen sink" is how it was described. - Systems architecture now comprises the majority of his job — primarily NetSuite-focused with Shopify integration through Celigo and the Craft web store for the Watchu brand - Mornings consumed by European time zone emails; 40% of day on sales-specific inquiries (permissions, accommodations, certifications, shipping/finance requests, custom data exports) - Financial work consumes 20–30% normally but 100% currently due to year-end cleanup. Fiscal year ends on calendar year; reporting completed by end of February using review engagement (not full audit) after 20-year relationship with accounting firm - Works most closely with: Lanny (incredible systems knowledge, very engaged with NetSuite), Byron (European sales, long tenure, remembers everything), Jennine (relatively new, handles repetitive tasks like payment matching) - Main concern about consultants: implementation capability matters more than ideas — "company already has good ideas but struggles with deployment due to bandwidth constraints" **2. His Systems World (15 min)** Start broad, go deep. Let him map it for you. - Walk me through what's connected to NetSuite today — Shopify, anything else? - What data flows in and out, and what's manual vs. automated? - What custom work have the NetSuite consultants done? What's still on the wish list? - What does NetSuite do well here that people don't realize? - What's your biggest frustration with how it's configured today? - Where does reporting break down — what can't you pull easily? - If someone asks for a new field or report, what does that process look like? - How does the $50k/license cost shape what you can and can't do? - Any tools you're supposed to use but don't? Workarounds you've built yourself? Notes: - NetSuite (core ERP): Heavily custom-configured. Recently migrated from REST to SOAP-based integration, which disabled enormous functionality and broke multiple integrations. Server lag forces Conor to work in 50+ concurrent tabs to avoid performance penalties. Many functions hard-coded to prevent automation (e.g., cannot mass delete records; billing automation previously available, now removed). - Shopify + Celigo: Works extremely well. Celigo is the industry-standard API connector for Shopify-to-NetSuite communication. Integration is "extremely tight with minimal maintenance" — only one three-week issue resolution in its history. Shopify handles all customer transactions, payment processing, and communication as source of truth. Backend currency CAD, primary sales USD, customer-facing supports USD/CAD/EUR/GBP with goal of 9 currencies. Auto-deposit functions handle daily batching with currency conversions. Fully automated tracking number export and order closure. - Craft/Watchu web store: ZERO integration with NetSuite. All orders manually entered by sales agents despite 80% being high-volume, low-value component work. Sachi and Mandi manage operations with external developer Michiko (not on staff, limiting update capability). Future integration platform planned but blocked by the REST-to-SOAP migration breaking tax functionality. - CPQ module: 85% complete in primary testing, delayed due to philosophical re-examination of major modules. Three components: inventory module (transfer, bin replenishment, accuracy), CPQ module (sales-facing, data entry quality), state manager module (addresses NetSuite's limited sales order states — currently only "pending approval" and "pending fulfillment," the latter a catchall that auto-allocates inventory even on orders that will be canceled). Estimated 50 additional hours on Conor's side to reach sandbox completion, then production deployment, user testing, and dual-system transition period before go-live. - Item groups architecture: Creates severe reporting blindspots. Item groups are not true inventory records but shorthand for sales to quickly enter fixture configurations. Poor import functionality, terrible reporting, hollow inventory records. Cannot track pendant color popularity across fixture types, cannot consolidate multi-location inventory, cannot trace components back to specific configurations. Financial reporting works; fixture-level popularity analysis is extremely difficult. - Pricing: Master Excel sheet maintained exclusively by Conor — historical tracking for all pendants and fixtures. NetSuite pricing is set up from Excel calculations. No one else maintains this. - Inventory complexity: Components (smallest sellable unit), kits (dynamically constructed at POS from separately stocked components), assemblies (pre-constructed and pre-stocked). Could theoretically reach millions of SKU combinations from color and component variants. - Consulting support: Prolacta (current) considered best in industry, building CPQ module. Can be on-site but extremely expensive (~$10,000 for a few days). Advanced Customer Care Service (ACS) was a disaster — initial consultant excellent but promoted out, followed by constant problems. **3. Workflows & Pain Points (10 min)** Now you have the landscape — explore where it breaks. - Walk me through the replacement order process — what gets captured today and what's missing? - Molly mentioned there's one dropdown with ~15 options and you can't log multiple issues per order — is that your understanding too? - Tyler said he manually categorizes the monthly replacement report line-by-line to find patterns — does that reporting flow through you? - What's the most repetitive request you get? - What's the biggest thing the business asks for that the current setup can't do? - People in different departments use different codes for the same product — has that caused data problems on your end? - Hardest thing to teach a new hire about your systems: Notes: - 40% of day consumed by ad-hoc sales requests: Permissions, accommodations, certifications, shipping/finance requests, custom data exports. Projects constantly bumped by urgent operational matters — needs separate capacity for project work versus day-to-day operations. - Item group architecture blindspots: Cannot trace back which item groups specific components were used in. Cannot report on popular colors across different fixture types using the same pendant. Competitiveness is based on high common component reuse, but system obscures which products contain shared components. - CPQ perpetually deferred: 85% complete but can't finish because operational fires keep consuming bandwidth. Estimated 50 hours remaining but those hours never materialize. - Avalara was an "unmitigated disaster": Attempted ~1–1.5 years ago for tax automation. US-company bias; failed completely for Canada operations. 30+ instances reaching out for support; forced through tier-one technician queue with 2-week escalation to level-three support. Finally abandoned due to wasting weeks on each support interaction. TaxJar had similar issues. - Forecasting nearly impossible: Pending fulfillment inventory allocation shows items at negative (e.g., -120 units of green pendant) because allocated to unreleased orders. State manager module aims to solve this. - Reporting workarounds: Saved searches set up for Randy (receives 1–2 reporting requests per week). Excel used as alternative for complex analysis since Conor can sort data faster in Excel than in NetSuite. - Tax complexity: California sales tax is the primary challenge — one of four states vociferously enforcing foreign sales tax after the Wayfair case (~2013). Nested state, city, and regional taxes with building-level variation. Canadian operations significantly simpler (single tax rate by province). - Product launch bottleneck: New product launches consistently become huge bottleneck for several months while building glass stock. Staged process from component codes to pricing to photography to SKU matching. **4. How Work Comes to Him (5 min)** Flows naturally from pain points — how does the org interact with his world? - When other departments need something from NetSuite, how does that request come to you? - Do people come with clear asks, or are you translating vague needs into system changes? - How do you find out about product or process changes that affect your systems? - Molly said she's been waiting ~2 months to meet with the NetSuite consultants about replacement reporting — what's been the blocker there? Notes: - Requests come via email, Slack, walk-ups — no formal intake process. Conor is constantly translating vague needs into system changes. Randy receives 1–2 reporting requests per week, delivered via email or direct NetSuite access. - Randy prefers not recording decisions online — creates a pain point where he sometimes changes his mind on decisions that were never formally documented. Speed expectations high; ideas-oriented. - Molly's NetSuite consultant delay confirmed: Conor's bandwidth is the blocker. Prolacta is available but Conor doesn't have time to scope and manage the engagement while handling day-to-day fires. - Knowledge concentration risk: Limited team depth on systems knowledge outside Conor, Lanny, and Byron. Jennine handles front-end/busy work; Michelle and Megan in shipping are excellent at getting orders out the door but don't research complex situations. **5. Aspirations & Closing (10 min)** Protect this time. This is where the gold is. - What should NOT change about how things work today? - Where do you see the biggest opportunity to use data or automation that nobody's acting on? - If you had more bandwidth, what would you build? - Any interest in AI tools — have you played with anything? - Anything I haven't asked that I should know? - Anything you're nervous about? Notes: - Interested in Fivetran + Databricks for a data warehouse solution — more forward-thinking on data architecture than expected. Databricks provides a data lakehouse with data activation layers for AI/ML capability. Noted that finance teams are increasingly using Python in Databricks instead of Excel for faster, repeatable processing. - Open to teardown-and-rebuild if demonstrably better: differs from Randy, who believes new equals better. Conor requires change to have a clear reason and cohesive benefit. Unwilling to accept change for change's sake or nifty features without functional purpose. - If he had more bandwidth: Finish CPQ module, build Craft integration, get Deutsche Bank and CIBC to export directly to NetSuite, Excel dump to automatic upload capability. - AI appetite low but not hostile: Currently not using AI tools at work despite exposure. Skeptical of personal efficiency gain given communication style (fires off 10-second one-line emails). Not against AI tools but needs to see tangible efficiency benefit. Cannot accept AI hallucination risk for financial data — requires 100% accuracy for financial information versus lower accuracy acceptable for sales data. Noted Oracle (NetSuite parent) pushing their own AI. - Shopify + Celigo should NOT change — it works. The tight integration is a bright spot. - Willing to be shadowed for an observation day given sensitivity agreements. Only sensitive times are salary discussions. Randy will need to confirm access but Conor expects approval. - Recommended next step: Sit with Conor and Jennine for a soup-to-nuts sale process overview to understand mandatory vs. optional records and overall process flow. **Post-Interview Capture** - What did Conor care about most? Getting bandwidth to finish CPQ and fix the Craft integration. These are the two projects he knows would move the needle but can't reach because operational fires consume his capacity. - Top 3 pain points: - Actual tools used: NetSuite, Shopify, Celigo, Excel (pricing master sheet), Craft/Watchu - Change readiness: Open — not Skeptic as initially assessed. Requires clear reason and cohesive benefit, but explicitly open to teardown-and-rebuild if demonstrably better. Differs from Randy who believes new equals better. - AI appetite (1-10): 3 — not using AI at work, skeptical of personal efficiency gain but not against it, zero tolerance for hallucination in financial data. Noted Oracle pushing their own AI integration. - Handoff readiness — could this person own ongoing system work? Yes — this is the long-term systems owner. 14 years of institutional knowledge. He IS the systems architecture at Bocci. The question isn't whether he can own it; it's whether he has bandwidth to execute alongside his finance and shipping roles. - Quick wins: Unblock Craft integration (80% manual order entry is massive waste), complete CPQ module (85% done), automate Jennine's payment matching (repetitive task that should be automatable) - Surprises: Interested in Fivetran/Databricks — more forward-thinking on data architecture than expected for someone who presents as skeptical. Avalara was an abandoned disaster (30+ support interactions, weeks wasted). Works in 50+ concurrent NetSuite tabs to dodge server lag. REST-to-SOAP migration is the root cause of many current integration failures. - Confirms or contradicts Molly/Tyler/Brian: Confirms Molly's NetSuite consultant delay — Conor's bandwidth is the blocker, not Prolacta's availability. Confirms Tyler's assessment of fragmented systems. Adds critical context that the REST-to-SOAP migration is the root cause of many integration failures. Confirms the general picture of knowledge concentrated in too few people. - Key quotes: Source transcript: NetSuite Systems Architecture Overview

Interviewee: Patrick (Paddy) Mulcahy — Industrial Designer, Product Development (Bocci Luminaire + 22 Systems) Date: 2026-02-25 | Duration: ~62 min | Interviewer: Connor Pundick **TL;DR** Most candid interview in the discovery. Paddy named things nobody else has named. 3DX is dead — confirmed and closed. Decision ownership is the real ProDev bottleneck. Information hoarding is driven by job insecurity, not malice. NetSuite access for ProDev would unlock real tooling decisions. He leaves for Milan mid-April for six months — follow up before he goes. **Key Findings** **1. 3DEXPERIENCE Is Functionally Abandoned** Paddy's verdict: "It's a failing system that's never going to work, and he's not adapting. Every system is you adapt to your environment. Not be forced into a thing." Brought up Daniel unprompted within 6 minutes. In practice: 3DX is used only at end of process to check files in. All real work lives in local folders. Hyperlinks break and drawings disappear. License costs prevent production from viewing PDFs. Built for a 200-person company, not a 4-person ProDev team. His proposed alternative: organized folders + PDF exports + naming matched to NetSuite codes. "She has all her links in a folder, that folder goes into a folder. That's all we need." Connor's read: Third independent confirmation (Jasmine, Aaron, Paddy) that 3DX isn't being used as intended. Daniel is the lone advocate — conflict of interest noted. 3DX question is closed for discovery. **2. Decision Ownership Is the Real Bottleneck** Power board project: four versions, design ready, but nobody will commit on CNC locally vs. China. Paddy goes Randy → Tyler → stall. The 22 Systems adjustable pendant went through tooling three times because upstream decisions weren't locked before tooling. Pattern: design progresses optimistically → decisions deferred → reality hits at tooling → expensive rework. "You have to have control of the process from the start to the finish." **3. Information Does Not Flow Between Departments** Cross-departmental communication only happens at crisis points. No mechanism for production to participate in product development before launch. The structural problem: leaders who are "the communicator between departments" rather than building communication infrastructure. "Their job is to create communication between departments. It's not to be the communicator between departments." **4. NetSuite Access Is Gatekept From ProDev** ProDev gets a monthly returns/failure list. No real-time visibility. His proposal: scrap 3DX, make SolidWorks output match NetSuite item codes. "That's how we order it anyway." Lana already manually uploads some PDFs to NetSuite — formalize it. **5. 22 Systems Moves Even Slower Than Bocci** Separated for apparent efficiency, running at "snail's pace." Three tooling iterations on the same product. No lean process engineer in the building. Adjustable 22 (132 SKUs → 7) is a good outcome but took too long. **6. The Quiet Part About Org Dynamics** "I think everyone's in fear of losing their job here... Impostor syndrome." People hold onto task ownership because letting go raises the question of what their role is. This is upstream of most communication failures across the engagement. **7. Paddy's Self-Assessment** Highly systematic thinker, communicates in a way that reads as scattered. Works best with clear directives. Previously head of product development — moved out due to "communication issues" (flag for Connor: who made this call? what were the issues?). **Tools Confirmed in Use** - SolidWorks (primary CAD) - Local folders (actual file storage) - Email (proof of record for supplier drawings) - 3DX (compliance check-in only) - No NetSuite access — monthly returns list only - OpenAI (personal subscription) **Quick Wins** - Folder structure + PDF naming matching NetSuite codes — no new software - NetSuite read-only access for ProDev - Decision log for ProDev (lightweight, not 3DX-heavy) - Suspend 3DX further rollout **Action Items** - Follow-up interview with Paddy before April 15 — Connor - Confirm 3DX actual usage across ProDev — Connor / Brian - Pull Lana's NetSuite PDF workflow as reference model - Clarify Paddy's exit from Head of ProDev role context **Key Quotes**

Date: February 11, 2026 Attendees: Molly Donnici, Tyler Archibald, Connor Source: Jamie AI transcript + Connor's Wispr Flow review notes **Key Takeaways (Connor's Review)** **1. 22 Systems Support Workflow** High volume of repetitive questions pointing people to existing documentation. Need a support workflow that enables self-service and reduces back-and-forth. A bot with all documentation at hand (RAG-powered) would be valuable for both internal teams and customers. - 50-100 daily support emails on busy days, mostly repetitive technical questions - Customers are middle-aged contractors often reluctant about complex technology - Questions fall into small number of categories with existing answers in documentation - Much of current support is cut-and-paste from existing responses - 5 years of historical support email data available for training automation - 22 Systems is ideal chatbot testing ground given standardized question patterns - Chatbot hesitation due to risk of angering already frustrated customers **2. Living Documentation & RAG Bot** Support team currently has to go to the website, find documentation, and pass it to people for internal use. Need a centralized, always-current documentation source. - Sales team currently reaches out to support to ask product questions - A bot with all documentation/RAG at its fingertips would let sales self-serve - FAQ section is outdated and was ignored during website relaunch - Michiko (external contractor) managing technical website rebuild including new FAQ area - Documentation is overly technical — could benefit from conversational tone matching email responses - No single place where all critical information lives and is easily accessible **3. Product Support Request Tagging (NetSuite) — Replacement Orders? ** When replacement parts need to be sent, the data captured in NetSuite is insufficient for pattern detection. Need signal-based reporting. - NetSuite replacement orders have only one dropdown field with ~15 vague options - Cannot record multiple issues per replacement order (e.g., "4 LEDs broken + shipping damage") - Free text descriptions don't align with actual categorization - Currently takes days of manual email searching to identify patterns (e.g., discovering breakage was color-specific) - Desired state: Telemetry-style reporting — "This batch for this product in this color has had 4 replacement orders in 3 months → flag it" - That feedback loop should tie directly back to manufacturing/production - NetSuite consultants meeting overdue by ~2 months — Molly has documentation prepared on desired reporting structure **4. Product Development Workflow & Change Tracking** Reinforces the #1 value-add identified across interviews. The product lifecycle lacks formal tracking and sign-offs. Current flow: Omer + Jay (Ideation, secretive phase) → Brian's Pro Dev team (Productionizing) → Iterative documentation → Multiple departments Who needs to receive info from Pro Dev: - Mandy (Designer) — builds spec sheets, currently guessing due to incomplete info - Molly (Support) — needs to know what changed to support customers - Julienne (Compliance) — handles all compliance/certification status - Design team for documentation and approvals Key problems: - No digital sign-off system — designer can get Randy's approval without notifying Brian (PM) - Changes to existing products go unrecorded - Part changes affecting 20+ products are typically missed - Randy and Omer make ad-hoc decisions in random conversations without documentation - Randy sometimes forgets decisions - Departments operate as islands with limited cross-team visibility Monday board concept discussed: - Single object tracks product from ideation through production - Multi-approver sign-offs at checkpoints - Tied to Slack notifications for all relevant parties - Key implementers: Brian (Pro Dev), Molly (documentation), Tyler (operations) - Challenge: team perception that added process slows them down **Company Structure Detail** **Two Business Lines Under One Organization** Both funnel through NetSuite for core operations. **Tech Stack Notes from Interview** - Hiver: Gmail overlay for support — custom fields, AI/automation options available but underutilized. Previous Zendesk attempt created more problems. - Shopify: Powers 22 Systems e-commerce only - 3D Experience/PDM: Poor search functionality, team divided on usage, must know exact product name to find info - CRM opportunity: Layer over Gmail/Hiver to enable BI reporting while maintaining personal support brand. Flow: Hiver/Gmail → CRM → NetSuite → Looker Studio dashboards - Monday.com: Mentioned as concept for product workflow tracker **Recommended Next Conversations (from Molly)** - Bryan (Brian Magennis) — Lead product developer/project manager - Daniel (Nikitiuk) — Designer, interested in systems, formerly systems consultant - Mandy (Chang) — Keeper of all documentation - Sachi (Sachie Sakai) — Keeper of website materials for Bocci side - Julian — Certification tracking, phones, support (on vacation starting Monday) - Connor (NetSuite admin) — NetSuite administrator, schedule consultants meeting - Andrew (Nguyen) — Production/manufacturing **Action Items** This interview strongly reinforces the #1 Value-Add (Change Management Process) and adds concrete detail about the product development workflow, support pain points, and NetSuite limitations that should feed into Phase 1 recommendations.

**Interview: Lana Dooling — General Manager** Date: 2026-02-10 Duration: ~60 min Source: Jamie transcript **Role & Context** - General Manager overseeing day-to-day operations - Reports to Randy Bishop - Coordinates across sales, finance, marketing, support, and operations - Day-to-day sales coordination flows through Lana even though Erin technically manages overall sales **Tools & Systems** - Tools used daily: NetSuite, Gmail, Hiver (support + direct sales) - Key systems identified: **Org Structure Walkthrough** **Leadership** - Randy = business/ops. Omer = creative director (works from home, 2 small kids, neither maintains office/desk) - Omer interfaces ONLY with product dev and glass production — not finance, sales, warehouse, or support **Product Development (under Brian)** - Team: Vince, Jasmine, Patrick, Aaron, Daniel (22 Systems exclusive, 3D Experience lead) - Julien = contract FTE on 22 system, handles all regulatory compliance (UL, CE, country-specific) **Sales (under Erin)** - Erin = Chief Brand Officer, London, ~20 years tenure, founded dealer network - 22 Systems: Mia (office) + Vico (Europe) - Direct sales: Sachi + Emily (share Hiver inbox; Emily started 3 months ago, will be primary CRM contact, currently cleaning Pinterest + researching AI design tools) - Dealer — US: Juju. Canada/temp: Aly (covering Ellie's mat leave until May, also PM). Europe: Byron (Berlin, since 2009), Laura (Berlin), Esther (Vienna) **Operations** - Tyler = operations manager (vendor negotiations, tech support, travels to Czech Republic/China) - Andrew = warehousing, assembly, storage - Lanny = inventory accountant, places all POs based on prior year analysis - Dermot = tactical procurement (glass shop needs, shelves) - James = order fulfillment, lead time responses to sales - Connor Graham = NetSuite systems/integrations, ~14-15 years tenure, originally accountant **Glass Production** - Jay = technical + personnel management, travels frequently (Czech Republic, Portland) - Felicia = administrative side (color inventory, prototyping docs, furnace maintenance, go-to when Jay travels) - China vendor relationships through principal contact Robbie (supplies hardware/components, subcontracts) **Support** - Molly = principal support (Hiver). Julien (French) = email support alongside Molly - Paul = chief installer (complex technical/electrical cases only) - Tyler, Aaron, Lana involved in support decisions; Aaron + Lana determine client offerings (free replacements vs charges) **Architecture (under Omer)** - Team: Catherine, Apollo, Kevin, Tyler W. All projects currently internal (Governor's Point — Randy's land development in Washington) **Design & Creative** - Mandy = graphics + specs (reports to Aaron London for design direction) - Dennon = layouts + project support with Omer, shifting to sales drawings - Daria = art director (reports to Aaron + Johannes, leads campaigns with external partners Max + Kiernan) - Johannes = director of communications, Berlin (ideas-focused, limited day-to-day office interaction) **Marketing** - Bronwyn = operational marketing support + office manager (Klaviyo, newsletters, travel, events, supplies) **⭐ #1 VALUE-ADD ALIGNMENT: Change Management** Both Lana and Tyler independently identified the same critical gap: Product changes cascade across 8+ teams with NO formal checklist, documentation system, or handoff. Decisions rely on verbal communication and individual memory. Lana's example: Stem series dimmer dial color change (brass to black option) — ripple effects across inventory, photography, specs, sales, dealers, NetSuite, website, timing. Small change created extensive coordination across multiple teams. Changes often forgotten by time new inventory arrives, causing teams to scramble. Tyler's example: Prototype dimension changed from 7.5" to 8" in backroom discussion between two people. Took 2+ months before others accidentally discovered the change. This is the #1 get for the whole company for process and product change updates. **Post-Interview Capture** - Top 3 pain points: - Actual tools used: NetSuite, Gmail, Hiver, Klaviyo - Change readiness: Open (medium willingness to change per Wispr notes) - Quick-win candidates: Product change checklist/documentation system, formalized handoff process for cross-departmental changes - Surprises / unexpected insights: Day-to-day sales coordination actually flows through Lana, not Erin. Connor Graham has been there 14-15 years. Emily Vender (3 months in) positioned as future CRM point person. Source transcript: Lana Dooling Interview - Company Org Structure Overview

**Interview: Emily Vender & Sachie Sakai — Direct Sales (Bocci)** Date: Feb 13, 2026 | Duration: ~45 min Format: Joint interview — both handle direct (non-dealer) order flow Source: Jamie transcript **1. Role & Context** - Sachie: 3 years at Bocci. Originally hired for website redesign, role expanded to managing website, online orders, and customer service. Background: 12 years in custom architectural lighting manufacturing. Previously handled 60–70 emails/day alone with no breaks. - Emily: New hire supporting Sachie. Works closely with sales and support teams. Two-person team now manages email volume with time freed for other projects. - Division of work: Shared Hiver inbox for the design team. Both handle inbound emails from electrical contractors, distributors, interior designers, and non-dealer parties. Sachie carries deeper product/design expertise; Emily is building fluency and leaning into automation/GPT tooling. - Works most closely with: Michiko (web developer), Mandy (website design/spreadsheet edits), Molly (dealer-related and technical inquiries, FAQ development), Juju/Chu-Chu (sales/dealer management). **2. The Order Pipeline: Craft → NetSuite** - How orders arrive: Customer purchases on Bocci website → Craft generates a PDF order form containing shipping/billing address, contact info, PO, date, order ID, taxes, and discounts. - Payment: Handled by Craft before reaching the design team. Payment processing happens in Craft, not NetSuite. - Manual entry into NetSuite: Sachie/Emily copy-paste from PDF into NetSuite to create the order. This generates a PI (Proforma Invoice) number that is different from the Craft order number. Both numbers are attached to every order. - Order modifications: Happen for shipping method changes, address updates, and color/cable/pendant changes after initial order. - Design-hold orders: Some fixtures (sculptural cable, copper) require design drawings before production. Customers sometimes miss the website note and order anyway → order goes on hold until drawings completed. Recent example: 28.7 copper fixture. - Automation opportunity: The entire process is manual copy-paste from PDF to NetSuite. Direct Craft→NetSuite integration would eliminate this bottleneck. Confirms Conor interview: zero Craft-NetSuite integration, 80% high-volume/low-value component work. Integration platform planned but blocked by REST-to-SOAP migration. **3. Tools & Systems** - Hiver: Email management with tracking and team handoff. Emily experiencing significant bugs — assigned emails show team emails, confusing filters, described as "very buggy" on her machine. - Craft: E-commerce platform for online orders and payments. - NetSuite: Inventory and order management, payment processing. - Google Sheets: Pricing, product information, lead time tracking with real-time updates. Central repository spreadsheet contains SKUs, titles, pricing, and all product data. Designed as middleman between original inventory and website, eventually intended to integrate with NetSuite and Craft. - Design software: Rhino (3D modeling), Grasshopper (parametric plugin), Illustrator, InDesign. - ChatGPT: Both use it for everything. Emily wants premium account to build custom GPTs. - Gemini: Tested for image generation (Nano Banana feature), follows instructions well. Testing for inserting fixtures into customer space photos. - Jace: AI platform for email drafting — tested before Emily arrived but had Hiver integration conflicts and didn't work. - Zendesk: Tried briefly, showed the value of CRM functionality (automatic linking of customer conversations) but didn't achieve full adoption because sales still used emails. - Unofficial workarounds: Sachie built email templates by piecing together past responses and repeat patterns. Searching past emails faster than re-typing. - Tools supposed to use but don't: No formal CRM — manual search for past customer threads. **4. Email Management & Response Workflow** - Volume: 60–70 emails/day before Emily joined. Now split between two. - Response target: Same-day. 1–5 min for simple requests, up to 10 min for complex ones (pulling installation instructions, website resources). - 70% of emails are straightforward: Pricing, lead times, spec requests — information already on the website. Customers don't navigate to find it. - 30% are complex: Customer-specific design requests, architectural drawings ("what can we do with this?"), custom solutions requiring dimensions/photos. - Three main request themes: (1) Pricing requests (biggest category), (2) Lead times, (3) Trade account requests. - Installation inquiries: Significant volume, particularly for refurbished/custom houses with complications. Individual circumstances require custom solutions. - International: Shipping restrictions exist (China, Russia) but no clear documented list of countries company doesn't ship to — information scattered or in head knowledge. - Template system: Hiver pricing templates with copy-paste from product pages. Sachie built templates from past responses and repeat patterns. **5. Data Quality & Handoffs** - Product info source of truth: Google Sheets updated in real-time by Sachie and Mandy. Manual upload to website by clicking button (not automatic). Same workflow for FAQ updates. - Multiple spreadsheets: Track pricing, lead times, weights, and package sizes. Lead times kept updated on website to reduce spreadsheet cross-referencing. - Order entry risk: Copy-paste from PDF creates error potential. Dual numbering system (Craft order # vs. NetSuite PI #) adds complexity. - New product launches: No specific launch communication process surfaced. Conor's master Excel sheet sets pricing; Sachie/Mandy update Google Sheets and website. **6. Communication & Coordination** - Sales/dealer tension: Company prefers direct orders (100% margin) over dealer orders (15–50% margin). Trade accounts give 15% discount to interior designers/contractors, keeping the sale in-house. Randy prefers direct but values dealer relationships. - Project ownership unclear: Risk of design team and sales competing for same customer. Tracking spreadsheet exists but not always updated. Usually resolved by email check-in ("Is this familiar?"). - Pricing arbitrage pain: Customer gets pricing from design team, goes to dealer, then wants to come back for better pricing. Some dealers (Future Perfect) match; others won't. Creates difficult situations. - Emily's challenge: Lacks intuition about which projects design team should take vs. sales. Has assigned to Juju then found Juju already had it. Needs quick real-time decision-making to respond within a day. - Collaboration: Molly handles dealer-related and technical inquiries + FAQ development. Mandy handles website design + spreadsheet collaboration. Juju manages sales/dealer; Sachie working with her on sample program improvement. **7. Website & FAQ Management (Sachie)** - FAQ section: Currently messy and outdated. Downloaded from old website without cleanup. Categories exist (by series, mounting type, parts, orders, accounts) but organization is inconsistent. Missing newer products. - In-progress fix: Moving FAQ to spreadsheet model (like product pages). Web developer building spreadsheet-to-FAQ download workflow. Would give Sachie and Molly control, efficiency, and coordination. Manual upload allows verification before going live. - Product pages: Well-designed with pricing, details, specs, and installation instructions. Customers rate website as easy to use but still prefer emailing for answers. **8. Emily’s Transition** - Emily transitioning to Marketing/CRM under Lana. Driving factor: CRM gap is widely felt and Emily has automation/GPT skills from prior role. - Unclear which parts of current workflow carry over vs. get handed off. - Documentation need before transition not explicitly addressed in interview. **9. AI & Automation Opportunities (from the team)** - Level 1 email automation: Route straightforward pricing/lead-time/spec requests to custom GPT connected to live Google Sheets data. Draft responses for human review. - Email template generation: AI tool that learns from existing emails and develops new templates over time. - Image generation for design visualization: Gemini for inserting fixtures into customer space photos. Could convert customer photos to line drawings for professional presentation. - Email filtering/categorization: Labeling system to see email type at a glance. Batch similar types (all pricing, then all projects) to reduce context switching. - CRM functionality: Link customer conversations and see history. Currently manual search to find previous threads. Harder now with two people; Sachie's memory previously compensated. - Emily's GPT experience: Built custom GPTs at previous job, experienced with setup. Wants premium ChatGPT account. **10. Aspirations & Closing** - What should NOT change: Brand tone and voice (critical, non-negotiable). Personal touch and human review in emails. AI-written emails must have human review to add personality and context. Not comfortable with fully automated responses without oversight. - What they'd change: Time spent on tedious straightforward emails. Want email labeling/categorization by type to reduce context switching. - Ideal day: Focus on creative, design-intensive projects. Want bespoke customer experiences in conference space — propose solutions, show 3D models and live edits with real-time collaboration. Sachie excited about Grasshopper algorithm for parametric pendant population (randomizes orientations so fixtures don't look identical). - Automation vision: Level 1 handled completely, Level 2 drafted for review, Level 3 fully hands-on. Not about disliking the work — clearing space for better work. **Open Loops & Next Steps** **Post-Interview Capture (fill within 30 min)** - What did Emily care about most? Automation and GPT tooling. She brought strong energy around building custom GPTs and automating the email workflow. Has real understanding of business problems — not just technically interested but sees the operational impact. Strong candidate to own CRM and automation projects post-engagement. - What did Sachie care about most? Getting to project work. She's frustrated being stuck on rote email responses when she could be doing bespoke 3D modeling (Rhino/Grasshopper), client consultations, and custom fixture proposals. The design work is clearly where her passion and skill set live. - Top 3 pain points: - Actual tools used: Hiver, Craft, NetSuite, Google Sheets, Rhino, Grasshopper, Illustrator, InDesign, ChatGPT, Gemini - Orders per day/week (get the number): Not quantified — get in follow-up or day-in-the-life observation - Avg time per order transcription (get the number): Not quantified — get in follow-up - Content/asset management pain (1-10): 6 — FAQ section is messy and outdated but product pages are well-designed. Spreadsheet-to-FAQ fix is in progress. - Change readiness — Emily: 🟢 Champion — actively pushing for change, has GPT-building experience, wants premium ChatGPT account, sees herself as an automation owner. - Change readiness — Sachie: 🟢 Champion — has been advocating for process improvements (built email templates, spreadsheet workflows), tested Jace AI proactively, wants to clear space for higher-value design work. - Quick-win candidates: Custom GPT for Level 1 emails connected to live Google Sheets (pricing/SKU/lead time data) — Emily can build this. Also: better email response process in Hiver (templates, categorization, labeling by type to batch work and reduce context switching).

Date: February 13, 2026 Interviewer: Connor Pundick Role: Designer / Documentation Keeper Department: Creative Location: Vancouver Office Source: Jamie Notes — Mandy Interview Bocci **Pre-Interview Context** From Stakeholder Directory & Prior Interviews: - Per Molly (Feb 11): "Keeper of all documentation" — builds spec sheets, marketing materials, product documentation - Frequently guessing on spec sheets because hardware release info from Pro Dev comes slowly and incompletely - Needs to be part of formal sign-off workflow to get accurate info - Key recipient in the product dev → documentation pipeline - Per Brian interview: ProDev lifecycle (P1→T3) has gaps in handoff documentation — Mandy sits downstream of this **1. Role & Day-to-Day** - Role: Designer covering marketing, website, and documentation. Primary focus is marketing materials, website content, and SOPs. - Typical week: Fluctuates with product launches. Currently high — 3 concurrent launches: 93 new lite, Matter shape (plinths system), and Milan launch at end of month. - Email volume: Most emails from Aaron relate to product lifecycle projects rather than one-off customer requests. - Image management: Manages all product images across systems, including uploads to Swivl (online database used by sales for client distribution). - Capacity: Insufficient hours in workday to complete all commitments. Owes work to multiple people (Juju and others). Non-urgent items consistently pushed back. Team is aware of capacity constraints. **2. Documentation & Content Ownership** Product Documentation Process: - Minimum one-week lead time per document - Creates 3D-based instructions from product dev files in InDesign - Collaborates with Paul (installation specialist) who validates real-world accuracy vs. theoretical design - Gets sign-off from Paul, Tyler, and product dev leads before translation - Limits detail to avoid multi-page docs due to translation costs Version Control Gap (Critical Finding): - Sales sells multiple versions (v2, v3, v4) simultaneously while Mandy is unaware of changes - Product dev doesn't know which version is being ordered or when new versions roll out - Lanny/supply chain determines switchover timing but this isn't automatically communicated - Tyler must explicitly notify Mandy when new versions ship so website can be updated - Has published documentation that became incorrect when product dev made post-handoff changes Storage: - Stores finished documentation on design server with archive folder system - Dates files to track versions (no formal version control system) - For small changes, sometimes overwrites old file; for small-medium changes, archives - Works alone managing server — manual system functions without conflicts - Molly occasionally retrieves older versions and knows where to find them **3. Image & Asset Management** Naming Convention: Project name → product name → photographer name → four-digit sequence number Tagging (Adobe Bridge): Product, color, location, photographer, year, distinguishing features. Enables sales team to search by tags in Swivl. Time Investment: Minimum 30 minutes per fixture for small projects; full day+ for large multi-fixture shoots. Varies by number of product types and locations. Two Categories: - Institute images (real-world settings) — retained across all periods, don't lose value - Studio photos (white background) — updated/replaced as newer versions created - Intentionally removes old studio photos from Swivl to prevent sales from accessing outdated imagery Storage: Both ShareBrowser (internal server, everything including older content) and Swivl (online cloud, primarily newer content since ~2028). New images uploaded to both simultaneously. **4. Marketing & Website Materials** Bocci Website: - Sets up new product pages using pre-built templates — no coding required - Copies template, fills in product-specific information, deploys at product launch - Manages all content updates: specs, version revisions, documentation - Emily handles lead time information and back-order status (not Mandy) 22 Systems Website (Top Pain Point): - Significantly more painful than Bocci site — built on Shopify with complex workarounds and multiple plugins - Must support multiple markets (grouped geographically) × 5 languages each (EN, FR, ES, DE, IT) - Single content update multiplied across all markets and languages; each market has different outlet requirements - Cannot verify translation quality — cut-and-paste from translations introduces unknowable errors - No complaints received yet, but Mandy aware issues likely exist - Blocks full days for new product launches — works from home without interruptions - Mia and Connor sometimes assist with translation work when overwhelmed - Uses RotoCon (half-AI translation software) — uploads files, assigns to human translators, then exports **5. Tools & Systems** 3D Experience Detail: Product dev often bypasses by packaging files locally and sending directly. 22 Systems team (particularly Julian) uploads files properly, making access more straightforward. Local file transfers frequently have broken links. Resolution varies from same-day to several days. Far less seamless than previous Autodesk experience despite using SOLIDWORKS. **6. Pro Dev → Documentation Pipeline** Information Flow Gaps: - Product information often arrives late or incomplete - Product dev sometimes indicates current version is active while next version with changes is in progress - Information gaps regarding which version is shipping - Ideal scenario (product finalized before documentation) never happens due to certification requirements - Has published documentation that became incorrect after post-handoff product dev changes What Mandy Needs: - Better version tracking so everyone knows which version is being sold - Wishes team had more time for product testing to reduce number of versions - Clear approval process — currently determines significance of changes herself based on visual changes or altered installation steps - Must ask Tyler directly to confirm which version is currently shipping Approval Gap: - No formal approval process for Bocci documentation or dedicated copywriter - Uses ChatGPT to edit copy, then Paul reviews for content/clarity - 22 Systems requires approval from Mia and Erin for marketing updates, but unclear who approves what **7. SOPs & Process Documentation** - No formal SOP repository or process emerged from the conversation - Documentation storage is file-date based on the design server with an archive folder system - Mandy works alone managing the server — the manual system functions because there are no conflicts - No version control system beyond filename dating **8. Pain Points & Time Sinks** #1 — 22 Systems Website Complexity: - Multiple languages × multiple markets = exponential update burden - Tedious manual translation cut-and-paste - Translator inconsistency creates quality issues she cannot catch

**Randy Bishop — Informal Conversation — 2026-02-18** Context: Hallway/informal conversation with Randy, not a formal interview. What Randy shared: Randy mentioned awareness of AI coding tools and their potential for the website side of the business. He was talking through the idea that AI-assisted development (tools like Claude Code, Claude co-work, etc.) could automate a lot of the website functionality work. Connor's read: This is a signal that leadership is already thinking about AI beyond just ChatGPT — Randy is aware of the coding automation space specifically. Worth exploring whether the website team has capacity and readiness for these tools. **Follow-Up** - Confirm with Michiko: Is the website team currently using any AI coding tools? - If not: Assess capacity for the team to adopt tools like Claude Code, Claude co-work for automating website development workflows - Broader implication: If Randy is already thinking about this, there may be appetite for a broader AI tooling rollout beyond just the ChatGPT business accounts that ProDev wants

**Daria Mikhailiuk — Art Director** Date: Feb 18, 2026 | Duration: ~30 min Reports to: Erin / Johannes (creative direction), Randy (budget) | Location: Vancouver Office Interviewer: Connor Pundick Context: Started August 2025. Manages digital assets and in-house creative production, replacing external agencies. Art Director title but functionally also producer, PM, and de facto marketing manager. **1. Role & Context** - Typical week: Full-scope creative production — pre-production through post. Art direction, styling, editing. Currently developing 5 simultaneous projects with uncertain sequencing. Two meetings per week with director. - How long at Bocci: Since August 2025 (~6 months) - Works most closely with: Fahim (primary in-house photographer, part-time), Kate (website/archival photography), Ally (PM + sales, stretched thin), Behi. Direction from Erin (PR firm in New York), Johannes, and Randy (budget sign-off). - What they enjoy most: Describes the environment as "a big playground that feels like freelancing under one roof." Values creative freedom and collaborative culture. - Notable: Has completed 6+ campaigns since August: three portable campaigns, one major business installation, artist collaboration, plus tests and in-situ photography. Does substantial PM work on top of creative — no formal marketing manager exists. Connor's read: Daria has a lot on her plate and is clearly impacted by the chaos on the ProDev side. She wants to maintain the open creative atmosphere but could use systems and process in the company. She's downstream of every unstructured product release. Notes: Jamie labeled Daria as "Speaker 1" throughout the transcript. Jamie also references "Amir" which is likely Omer Arbel. **2. Creative Workflow & Assets** **Campaign Development Process** - No formal briefs — receives vague or informal direction, sometimes indirectly through Erin's PR firm with specific asset requests for magazine placements - Creates deck of up to 5 ideas → presents to Erin and Johannes → Randy for budget sign-off → pre-production (casting, coordination with Fahim) → shoot → post-production (1-1.5 weeks) - Turnaround: typically 3 weeks start to finish. Ideal: 5 weeks. Sometimes compressed to 2 weeks or even 1 week if rushed. - For product launches: needs minimum 3-week lead time before launch date - Prefers explicit deadline for ad assets rather than product launch date **Asset Management** - Assets sent to Mendy for storage in Swivel (primary DAM/storage system) - Before Mendy, system was disorganized with unclear asset locations - Daria finds Swivel easy to use with helpful categories and searchable archives - For press/magazine: assets delivered to Erin who distributes further **3. Communication & Decisions** **Product Information Problem (Her #1 Issue)** - Product info is scattered and inconsistent — existing products are documented and searchable online, but new products have conflicting information - 93 campaign case study: Required 3 reshoot iterations because different product versions kept arriving. Wasted resources and money. - 141 plane for Milan: Conflicting guidance on readiness for shoot vs product not being ready. No single person has complete information. Smaller version being developed with uncertain timing. Displayed in Milan, New York, and Paris but unclear which venue uses which version. - Information gathered by: walking around, overhearing conversations, inspecting prototypes, informal discussions across departments - No proactive communication system — must actively investigate and collect info piece by piece - Does not have one-on-ones with Omer due to his busy schedule — bases understanding through others and observations Connor's read: This is the ProDev chaos cascading directly into the creative department's budget and timeline. When product changes happen without structured process, Daria's team is one of the first to absorb the cost — literally, through reshoots and scrapped work. **What She Wants** - Centralized system showing product development stage — statuses like "in development," "hardware approved," "Omer approved" - Regular reports or summaries about company initiatives, Milan preparations, exhibitions - Weekly update: 1 sentence per person on what they're working on - Earlier glimpses of product iterations so she can start conceptualizing before final versions **4. Tools & Systems** - Tools used daily: Adobe suite, ChatGPT, Canva (decks), Figma, Final Cut Pro (video editing), Gmail - Tools supposed to use but don't: None mentioned - Workarounds built yourself: Ad hoc information gathering through walking around and overhearing conversations - Magic wand fix: Product status dashboard - AI tool usage: ChatGPT ~50% of days — emails, briefs, schedules. Paid personal version. Also uses Adobe's built-in plugins for image generation. Open to exploring other AI platforms but hasn't tested competitors. **5. Team & PM Bottleneck** - Ally handles both project management and sales across ALL departments — stretched extremely thin - Daria does heavy producing/PM work on top of creative - Ideally: dedicated PM per department instead of single person overseeing all - Daria excited for Ally to focus more on PM **6. Aspirations & Closing** - What should NOT change: The warm, artistic, collaborative environment. "Too much regulation would take the soul out of work." Appreciates that process is somewhat chaotic because it preserves the artistic environment. - If she could fix one thing: Product status visibility — a dashboard or regular updates so she knows what's coming and can plan ahead - Nervous about: Inefficiency and wasted resources when information gaps lead to reshoot iterations or scrapped work. Feels personally responsible for waste even though she acknowledges she shouldn't. **Post-Interview Capture** - What did Daria care about most? Knowing what's happening with products so she can do her job without wasting resources on reshoots and scrapped work. - Top 3 pain points: - Actual tools used: Adobe suite, ChatGPT (~50% of days), Canva, Figma, Final Cut Pro, Gmail, Swivel (DAM via Mendy) - Change readiness: Open — wants structure but protective of creative culture. Explicitly concerned about over-regulation killing the artistic environment. - Quick wins: - Surprises: The 93 campaign requiring 3 reshoots is a concrete, costly example of the information flow problem. Also that she has no one-on-ones with Omer and bases her creative direction on secondhand observations. - Confirms or contradicts prior interviews: - Key quotes: - New stakeholders identified: Source transcript: Daria Mikhiliuk -- Interview (Jamie) — Note: Jamie labeled Daria as "Speaker 1" and references "Amir" (likely Omer Arbel)

**Interview: Emily Vender & Sachie Sakai — Direct Sales (Bocci)** Date: Feb 13, 2026 | Duration: ~45 min Format: Joint interview — both handle direct (non-dealer) order flow Source: Jamie transcript **1. Role & Context** - Sachie: 3 years at Bocci. Originally hired for website redesign, role expanded to managing website, online orders, and customer service. Background: 12 years in custom architectural lighting manufacturing. Previously handled 60–70 emails/day alone with no breaks. - Emily: New hire supporting Sachie. Works closely with sales and support teams. Two-person team now manages email volume with time freed for other projects. - Division of work: Shared Hiver inbox for the design team. Both handle inbound emails from electrical contractors, distributors, interior designers, and non-dealer parties. Sachie carries deeper product/design expertise; Emily is building fluency and leaning into automation/GPT tooling. - Works most closely with: Michiko (web developer), Mandy (website design/spreadsheet edits), Molly (dealer-related and technical inquiries, FAQ development), Juju/Chu-Chu (sales/dealer management). **2. The Order Pipeline: Craft → NetSuite** - How orders arrive: Customer purchases on Bocci website → Craft generates a PDF order form containing shipping/billing address, contact info, PO, date, order ID, taxes, and discounts. - Payment: Handled by Craft before reaching the design team. Payment processing happens in Craft, not NetSuite. - Manual entry into NetSuite: Sachie/Emily copy-paste from PDF into NetSuite to create the order. This generates a PI (Proforma Invoice) number that is different from the Craft order number. Both numbers are attached to every order. - Order modifications: Happen for shipping method changes, address updates, and color/cable/pendant changes after initial order. - Design-hold orders: Some fixtures (sculptural cable, copper) require design drawings before production. Customers sometimes miss the website note and order anyway → order goes on hold until drawings completed. Recent example: 28.7 copper fixture. - Automation opportunity: The entire process is manual copy-paste from PDF to NetSuite. Direct Craft→NetSuite integration would eliminate this bottleneck. Confirms Conor interview: zero Craft-NetSuite integration, 80% high-volume/low-value component work. Integration platform planned but blocked by REST-to-SOAP migration. **3. Tools & Systems** - Hiver: Email management with tracking and team handoff. Emily experiencing significant bugs — assigned emails show team emails, confusing filters, described as "very buggy" on her machine. - Craft: E-commerce platform for online orders and payments. - NetSuite: Inventory and order management, payment processing. - Google Sheets: Pricing, product information, lead time tracking with real-time updates. Central repository spreadsheet contains SKUs, titles, pricing, and all product data. Designed as middleman between original inventory and website, eventually intended to integrate with NetSuite and Craft. - Design software: Rhino (3D modeling), Grasshopper (parametric plugin), Illustrator, InDesign. - ChatGPT: Both use it for everything. Emily wants premium account to build custom GPTs. - Gemini: Tested for image generation (Nano Banana feature), follows instructions well. Testing for inserting fixtures into customer space photos. - Jace: AI platform for email drafting — tested before Emily arrived but had Hiver integration conflicts and didn't work. - Zendesk: Tried briefly, showed the value of CRM functionality (automatic linking of customer conversations) but didn't achieve full adoption because sales still used emails. - Unofficial workarounds: Sachie built email templates by piecing together past responses and repeat patterns. Searching past emails faster than re-typing. - Tools supposed to use but don't: No formal CRM — manual search for past customer threads. **4. Email Management & Response Workflow** - Volume: 60–70 emails/day before Emily joined. Now split between two. - Response target: Same-day. 1–5 min for simple requests, up to 10 min for complex ones (pulling installation instructions, website resources). - 70% of emails are straightforward: Pricing, lead times, spec requests — information already on the website. Customers don't navigate to find it. - 30% are complex: Customer-specific design requests, architectural drawings ("what can we do with this?"), custom solutions requiring dimensions/photos. - Three main request themes: (1) Pricing requests (biggest category), (2) Lead times, (3) Trade account requests. - Installation inquiries: Significant volume, particularly for refurbished/custom houses with complications. Individual circumstances require custom solutions. - International: Shipping restrictions exist (China, Russia) but no clear documented list of countries company doesn't ship to — information scattered or in head knowledge. - Template system: Hiver pricing templates with copy-paste from product pages. Sachie built templates from past responses and repeat patterns. **5. Data Quality & Handoffs** - Product info source of truth: Google Sheets updated in real-time by Sachie and Mandy. Manual upload to website by clicking button (not automatic). Same workflow for FAQ updates. - Multiple spreadsheets: Track pricing, lead times, weights, and package sizes. Lead times kept updated on website to reduce spreadsheet cross-referencing. - Order entry risk: Copy-paste from PDF creates error potential. Dual numbering system (Craft order # vs. NetSuite PI #) adds complexity. - New product launches: No specific launch communication process surfaced. Conor's master Excel sheet sets pricing; Sachie/Mandy update Google Sheets and website. **6. Communication & Coordination** - Sales/dealer tension: Company prefers direct orders (100% margin) over dealer orders (15–50% margin). Trade accounts give 15% discount to interior designers/contractors, keeping the sale in-house. Randy prefers direct but values dealer relationships. - Project ownership unclear: Risk of design team and sales competing for same customer. Tracking spreadsheet exists but not always updated. Usually resolved by email check-in ("Is this familiar?"). - Pricing arbitrage pain: Customer gets pricing from design team, goes to dealer, then wants to come back for better pricing. Some dealers (Future Perfect) match; others won't. Creates difficult situations. - Emily's challenge: Lacks intuition about which projects design team should take vs. sales. Has assigned to Juju then found Juju already had it. Needs quick real-time decision-making to respond within a day. - Collaboration: Molly handles dealer-related and technical inquiries + FAQ development. Mandy handles website design + spreadsheet collaboration. Juju manages sales/dealer; Sachie working with her on sample program improvement. **7. Website & FAQ Management (Sachie)** - FAQ section: Currently messy and outdated. Downloaded from old website without cleanup. Categories exist (by series, mounting type, parts, orders, accounts) but organization is inconsistent. Missing newer products. - In-progress fix: Moving FAQ to spreadsheet model (like product pages). Web developer building spreadsheet-to-FAQ download workflow. Would give Sachie and Molly control, efficiency, and coordination. Manual upload allows verification before going live. - Product pages: Well-designed with pricing, details, specs, and installation instructions. Customers rate website as easy to use but still prefer emailing for answers. **8. Emily’s Transition** - Emily transitioning to Marketing/CRM under Lana. Driving factor: CRM gap is widely felt and Emily has automation/GPT skills from prior role. - Unclear which parts of current workflow carry over vs. get handed off. - Documentation need before transition not explicitly addressed in interview. **9. AI & Automation Opportunities (from the team)** - Level 1 email automation: Route straightforward pricing/lead-time/spec requests to custom GPT connected to live Google Sheets data. Draft responses for human review. - Email template generation: AI tool that learns from existing emails and develops new templates over time. - Image generation for design visualization: Gemini for inserting fixtures into customer space photos. Could convert customer photos to line drawings for professional presentation. - Email filtering/categorization: Labeling system to see email type at a glance. Batch similar types (all pricing, then all projects) to reduce context switching. - CRM functionality: Link customer conversations and see history. Currently manual search to find previous threads. Harder now with two people; Sachie's memory previously compensated. - Emily's GPT experience: Built custom GPTs at previous job, experienced with setup. Wants premium ChatGPT account. **10. Aspirations & Closing** - What should NOT change: Brand tone and voice (critical, non-negotiable). Personal touch and human review in emails. AI-written emails must have human review to add personality and context. Not comfortable with fully automated responses without oversight. - What they'd change: Time spent on tedious straightforward emails. Want email labeling/categorization by type to reduce context switching. - Ideal day: Focus on creative, design-intensive projects. Want bespoke customer experiences in conference space — propose solutions, show 3D models and live edits with real-time collaboration. Sachie excited about Grasshopper algorithm for parametric pendant population (randomizes orientations so fixtures don't look identical). - Automation vision: Level 1 handled completely, Level 2 drafted for review, Level 3 fully hands-on. Not about disliking the work — clearing space for better work. **Open Loops & Next Steps** **Post-Interview Capture (fill within 30 min)** - What did Emily care about most? Automation and GPT tooling. She brought strong energy around building custom GPTs and automating the email workflow. Has real understanding of business problems — not just technically interested but sees the operational impact. Strong candidate to own CRM and automation projects post-engagement. - What did Sachie care about most? Getting to project work. She's frustrated being stuck on rote email responses when she could be doing bespoke 3D modeling (Rhino/Grasshopper), client consultations, and custom fixture proposals. The design work is clearly where her passion and skill set live. - Top 3 pain points: - Actual tools used: Hiver, Craft, NetSuite, Google Sheets, Rhino, Grasshopper, Illustrator, InDesign, ChatGPT, Gemini - Orders per day/week (get the number): Not quantified — get in follow-up or day-in-the-life observation - Avg time per order transcription (get the number): Not quantified — get in follow-up - Content/asset management pain (1-10): 6 — FAQ section is messy and outdated but product pages are well-designed. Spreadsheet-to-FAQ fix is in progress. - Change readiness — Emily: 🟢 Champion — actively pushing for change, has GPT-building experience, wants premium ChatGPT account, sees herself as an automation owner. - Change readiness — Sachie: 🟢 Champion — has been advocating for process improvements (built email templates, spreadsheet workflows), tested Jace AI proactively, wants to clear space for higher-value design work. - Quick-win candidates: Custom GPT for Level 1 emails connected to live Google Sheets (pricing/SKU/lead time data) — Emily can build this. Also: better email response process in Hiver (templates, categorization, labeling by type to batch work and reduce context switching).

**Chu-Chu — Sales, US Dealers** Date: Feb 12, 2026 | Duration: 45 min **1. Role & Context (5 min)** - How long at Bocci: 13 - Typical day/week: - How many dealers do you manage: - Who do you work most closely with internally: - Most important information sources ** Goal — more powerful image search - data input **2. Sales Process & Dealer Relationships (15 min)** Let her walk you through her world before you dig into systems. - Walk me through what happens from when a dealer reaches out to when an order is placed. - What does a typical dealer need from you that they can't get on their own? - What questions do dealers ask you over and over? - When a dealer asks you something you don't know, where do you go to find the answer? How long does that take? - How do you keep track of your dealers — conversations, orders, follow-ups? - What does the handoff look like once an order is placed — who takes it from there? - How do you find out about new products, pricing changes, or spec updates? Notes: **3. Tools & Systems (10 min)** - Tools used daily: Mac Mail , Netsuite, Chrome, Slack, Spreadsheets for tracking. - Tools supposed to use but don't: - Workarounds you've built yourself: - Do you use a CRM or is it email/spreadsheet-based? - Do you touch NetSuite at all? If so, for what? - How do you pull product info when a dealer needs specs, lead times, or availability? - AI tool usage: Notes: **4. Communication & Pain Points (5 min)** - How do you find out about product changes that affect what you tell dealers? - Has there been a time a dealer got wrong info — pricing, specs, availability — because something changed and you didn't know? - What's the hardest part of your job that nobody sees? Notes: **5. Aspirations & Closing (10 min)** - What should NOT change about how sales works today? - If you could fix one thing about your day, what would it be? - What do dealers complain about most? - Anything I haven't asked that I should know? - Anything you're nervous about? Notes: **Post-Interview Capture** - What did Chu-Chu care about most? - Top 3 pain points: - Actual tools used: - Change readiness: Champion / Open / Neutral / Skeptic - Quick wins: - Surprises: - Confirms or contradicts Molly/Tyler/Brian: - Key quotes: Pain : - Version control - - Usage : - Images: Lana knows everything Erin wants to vet for coolness Johannes - he is well connected — seeker

Date: 2026-03-02 | Duration: ~90 min | Interviewer: Connor Pundick Role: Inventory Control / Purchasing | Department: Production / Finance Reports to: Tyler Archibald Jamie transcript: Inventory Management Deep Dive: Bacci **Framing** Lani has been at Bocci nearly 10 years (with an 18-month gap 2022–2024) and is the primary person bridging the physical world of inventory — containers from China, Czech glass, Surrey warehouse — with NetSuite's financial record. She manages all external sourcing, intercompany transfers, and the monthly PO analysis process that drives purchasing decisions. She was flagged in advance as analog-resistant and unlikely to embrace change. Connor's read: that characterization is wrong. Connor's read: Lani is trust-before-adoption, not change-resistant. She's spent 10 years building accuracy into a complex, manual-heavy environment and isn't going to abandon that instinct lightly. But she opened up significantly in the final 20–30 minutes of the call — more excited, more candid — once she felt like Connor would actually invest in her, not just automate around her. The shift happened around the AI conversation and the offer of training and support. Nobody has invested in teaching her properly. That's the lever. **Key Findings** **1. The Misread: Not Analog, Trust-First** Lani was described pre-interview as analog and change-resistant. That framing doesn't hold up. She's self-taught on every tool she uses — NetSuite, Excel, barcode systems — and is genuinely curious about AI. What looks like resistance is actually a validation requirement: she wants to run old and new systems in parallel until the new one earns her trust. This is how she's maintained data accuracy for a decade in a high-complexity, low-documentation environment. It's a feature, not a bug. Connor's read: She explicitly said she learned everything on the job with little external support. Training and guidance lit her up. That's the entry point. **2. Prolecto Is Closing the NetSuite Gaps — C22's Opening Is What's Outside** Prolecto (NetSuite consultants) has active projects underway: - Purchasing dashboard (replacing her 7-export manual spreadsheet — launching end of month) - Receiving system update (Thursday) - Long-term pallet tag automation (Surrey warehouse) - CPQ system (sandbox, active testing) These are all NetSuite-internal. Connor's read: Prolecto builds the pipes. C22's opening is everything outside those walls — the analysis layer, AI-assisted judgment, cross-system intelligence, and the human-flexibility that rule-based NetSuite flows won't provide. Specific C22 opportunities she surfaced: - Analysis layer: Database-style monthly PO analysis that auto-flags anomalies vs. prior month, rather than 1.5 days of manual manipulation - AI training with Claude: She expressed openness; Connor committed to running her through this when they move forward - Excel / count sheet automation: She asked about faster ways to build physical inventory count sheets - Validation tooling: She needs ability to run old and new reports side-by-side to trust transitions **3. 1.5 Days/Month of Manual Analysis — The Direct Target** Every month Lani runs 7 separate NetSuite reports, manually combines them in Excel, and builds a color-coded PO analysis spreadsheet (pink = <18 months, orange = <12 months, yellow = <8 months). This drives all purchasing decisions. The purchasing dashboard will automate the data pulls. It will not replace the judgment layer — knowing which POs to flag, which anomalies need investigation, what seasonal fluctuations mean for the next order. That's where C22 adds value. She tracks two sales models simultaneously (last 12 months actual vs. last 4 months × 3) and uses the larger number as a buffer for seasonal swings. This logic needs to be preserved in any replacement system. **4. Surrey Warehouse Autonomy — A Phased, Blocked Project** Surrey is ~3x the size of Vancouver with more storage capacity but zero barcode scanning. The plan to enable Surrey-led picking and order fulfillment is: - Pallet tag automation (Prolecto building now) - Barcode rollout at Surrey (post-Vancouver stabilization) - Lexi identified as the right person to lead Surrey inventory - Eventually: a third inventory person in Vancouver, Surrey does its own builds and picking Currently blocked on hiring decisions by Andrew/Tyler. No timeline defined. Lani is the bottleneck by design until the system is ready — and she knows it. The 22s are the first product candidate for Surrey-based picking once barcodes are live. **5. Institutional Knowledge as a Liability — and an Opportunity** 10 years at Bocci means Lani is the memory for vendor relationships, product history, cost calculations, and why certain workflows exist. She spends ~1–1.5 extra hours per week digging through old emails to answer historical questions. She confirmed the "Randy Warehouse" virtual location — a NetSuite construct for items Randy pulls off shelves without proper ordering. Confirms the Founder's Trap pattern (Theme D) from prior interviews. C22 opportunity: Knowledge capture and surfacing. If we build the Product Ontology Layer, Lani's institutional knowledge is a primary input. She's also a candidate for a role-specific AI tool that can surface historical context without manual email archaeology. **6. Data Integrity Issues Are Active, Not Theoretical** Several specific pain points with real operational consequences: - NetSuite partial fulfillment save bug: Creates temporary VINs in the barcode system, causes data integrity issues (confirmed from Dermot's interview) - Bins system rigidity: Once items are received into bins, 10+ linked transactions make corrections nearly impossible. Dermot uses a "Variance Warehouse" to hold problem items; Lani reconciles weekly - OzLink issues: Barcode scanning provider is having significant problems; RF-SMART being evaluated as replacement - Czech Republic glass vendor for 141s: Doubled pricing due to breakage/rework — Tyler negotiating per-piece rates or alternative supplier. This is active and unresolved. **7. CPQ — Lani Is a Key Testing Partner** The CPQ system (22 Systems configurator, ~5 years in development, active build since 2022) is currently in sandbox. Lani and Connor are the ones testing question-tree logic and validating that correct parts populate on work orders. The 22 Systems product is also changing "very soon" — removable faceplate and barrel design — which means stock targets are dropping from 18 to 12 months. Timing TBD. **8. Change Readiness** 🔵 Open — with a clear design requirement: show her, don't just tell her. She needs to see old and new side-by-side. Training and support are the relationship builders here. She's not a skeptic; she's a validator. **Contradictions / Corrections to Prior Information** - Pre-interview characterization as analog/resistant = wrong. She's cautious and evidence-based, not change-averse. The person who said this may have seen her push back on something untested and misread it. - Reports To: Pre-interview data said Andrew Nguyen. She said she reports to Tyler. Update confirmed. **Action Items** **Themes Validated** - Theme A (Disconnected Lifecycles): Confirmed — no real-time visibility into glass shop production; production data flows manually via paper sheets to Dermot daily - Theme B (Manual Drag): Confirmed — 1.5 days/month on PO analysis; Surrey pallet management entirely in Google Sheets; multiple redundant data entry points - Theme D (Founder's Trap): Confirmed — "Randy Warehouse" virtual NetSuite location exists for items Randy pulls without proper ordering - Dermot's findings: Confirmed — barcode/bins data integrity issues, NetSuite partial fulfillment bug, Elements Glass loop **Watch For** - When Prolecto ships the purchasing dashboard — that's the moment to position the analysis layer conversation - Lexi at Surrey: understand her technical readiness before recommending Surrey autonomy timeline - Czech glass vendor situation for 141s — this could affect stock planning meaningfully if renegotiation fails - CPQ go-live: Lani is a test partner and needs to stay in the loop on timeline

**Interview: Andrew Nguyen — Production Lead** Date: 2026-02-10 Duration: ~45 min Source: Jamie transcript **Role & Context** - Production Lead managing warehousing, assembly, and storage - Reports to Tyler Archibald - Came from ISO 13485 certified medical device environment with formal documentation standards - Oversees ~30-person team split across production (5-7), order fulfillment (5-6), and warehousing (2-4), with 7 in mixed roles **Facility & Production** **Layout & Renovation** - Hot shop + cold shop currently congested. Renovation plans to move both upstairs, freeing downstairs for production expansion. **Process Flow** - Hot shop (glass blowing/forming/shaping) → Cold shop (post-processing) → Production floor (match glass with hardware from China, ~1,000 units per order) → QC at multiple stages → Final assembly + packaging (order fulfillment team, last hands before shipping) - Products use modular components for varying glass shapes/sizes - Upstairs team specifically QCs 28s for hardware compatibility despite manufacturing variations **Product Demand** - 28s: Highest value, currently top seller - 14s: Second highest value - 141: Released Feb last year, now highest demand with most glass committed to orders **Distribution & Sales Insights** - Europe disproportionately dominates: 20 dealers in France/Switzerland vs 20 across entire Western US - Sales team allocation reflects this: one US rep, one Canada rep, multiple EU reps (Aaron London, Kyra Germany, Viko, Esther) - Shipping completely in-house managed. FedEx primary carrier with preferred rate agreement. **Tools & Systems** - NetSuite: Functional but underutilized, operating in basic mode. Prolecta (external consulting) hired to optimize but will NOT address inventory scanning/barcode. - Inventory: Unique location codes adopted past 2 years, not fully realized. No barcode/scanning capability — manual entry risks human error. Chinese manufacturing partners willing to add barcodes if given specs, but company hasn't instructed them yet. - Communication: Texts (urgent), email (important), Slack (growing but not universal — order fulfillment team of 7 shares ONE account at one station), Google Drive (not universally accessible due to license costs) **Surrey Warehouse** - ~20x larger than main facility storage - Holds significant inventory: 5 years for hard-to-source glass, 12-16 months standard for most items, 3 months for some - Total inventory value difficult to determine without manual compilation across multiple sources **Documentation & Change Management** - SOPs difficult to maintain due to constantly changing details through informal conversations - Small cross-departmental pairs (Tyler/Randy, Tyler/Brian) make decisions without formal documentation - Example: Project 93 photography shoot required last-minute hardware sourcing, prompting immediate design iteration (double-sided tape to magnets) - Andrew proposed change logs similar to software versioning (e.g., 1.4.2) allowing self-update access - T1/T2/T3 designations already used for prototype stages indicating manufacturing readiness **Post-Interview Capture** - Top 3 pain points: - Actual tools used: NetSuite (basic), texts, email, Slack (limited), manual inventory tracking - Content/asset management pain (1-10): 6 - Change readiness: Open (came from ISO-certified environment, understands value of formal processes) - Quick-win candidates: Barcode scanning implementation (Chinese partners willing), change-log versioning system, Slack adoption for production team - Surprises / unexpected insights: 7-person order fulfillment team shares ONE Slack account. Surrey warehouse holds 5 years of hard-to-source glass. Andrew's medical device background means he deeply understands documentation value. Prolecta engagement won't touch inventory scanning. Source transcript: Andrew Nguyen Interview: Facility Operations and Process Overview **Follow-Up Interview — Feb 19, 2026** Date: Feb 19, 2026 | Duration: Context: Second interview. Since first conversation, have completed interviews with Brian (Product Dev), Daniel (22 System Design Lead), Tyler, Molly, Mandy, Emily/Sachie, Chu-Chu, and others. Looking to validate patterns and go deeper on production-specific workflows. **1. Since Last Time** - Last time we talked about barcode scanning and the Chinese partners being willing to add barcodes. Has anything moved on that? - You mentioned Prolecta was optimizing NetSuite but inventory scanning was out of scope. How's that engagement going? Anything changed? - Any changes to the Slack situation with the fulfillment team sharing one account? Notes: Not explicitly revisited in this session — conversation jumped straight into current operations and production metrics. Follow-up items from first interview remain open. **2. Design → Production Handoff** - When a design change comes through from Brian's team or Daniel on 22 System, how does that actually land on your desk? - What does a good handoff look like vs. a bad one? Walk me through a recent example of each. - Brian mentioned T1/T2/T3 stages — from your side, is that system actually working or is it more aspirational? - How often do you get hit with changes that weren't communicated — someone just shows up with a new spec? Notes: **3. Documentation & Version Control** - Last time you proposed change logs similar to software versioning. Has there been any traction on that idea? - Daniel is pushing 3D Experience for revision control on the design side — does that touch production at all? Would it help you? - When a product changes mid-production or between runs, how do you currently track what version you're building? Notes: **4. Cross-Team Dependencies** - Which teams create the most friction or confusion for your team day-to-day? - How does information flow between production and sales? Do you ever get blindsided by orders or commitments the sales team made? - What's your relationship with the Surrey warehouse in terms of coordination — who decides what moves when? Notes: **5. What Would Help** - If I could fix one thing for your team in the next 90 days, what should it be? - What's working well that we should protect and not mess with? - Anything you've been thinking about since our last conversation that you want to flag? Notes: **Post-Follow-Up Capture** - What changed since first interview? Andrew's team doubled daily capacity from 200-300 to 400+ orders with same headcount. He's now tracking daily metrics in a spreadsheet (took a year to implement). More confident and detailed about production operations. - New pain points surfaced: - Confirms or contradicts other interviews: - Updated change readiness: Champion (upgraded from Open). Andrew is data-driven, process-oriented from ISO background, wants to "run this operation surgically" but frames it collaboratively, not as a mandate. - Key quotes: (from Jamie transcript) - Connor's read: Really good chat. Andrew has great ideas and is very data-driven. Thinks others could benefit from that discipline. He didn't seem like he was holding back but also doesn't think process formalization should necessarily come from him — sees it as a growth opportunity for the org. Positive call, got cut short. Andrew recommended talking to Lani ("old school, paper is protection"), Dermot ("new school, good balance"), and Jacob ("sense of what goes on downstairs"). His aspiration: "could run this business in a way that would feel clean, every operation as clean and economical as possible — surgically." Framed through ISO-lite: take the positive things from formal standards without the bureaucracy. Source transcript: Production Operations and Living Space Fix **Standard Interview Questions (Production)**

Date: February 11, 2026 Attendees: Molly Donnici, Tyler Archibald, Connor Source: Jamie AI transcript + Connor's Wispr Flow review notes **Key Takeaways (Connor's Review)** **1. 22 Systems Support Workflow** High volume of repetitive questions pointing people to existing documentation. Need a support workflow that enables self-service and reduces back-and-forth. A bot with all documentation at hand (RAG-powered) would be valuable for both internal teams and customers. - 50-100 daily support emails on busy days, mostly repetitive technical questions - Customers are middle-aged contractors often reluctant about complex technology - Questions fall into small number of categories with existing answers in documentation - Much of current support is cut-and-paste from existing responses - 5 years of historical support email data available for training automation - 22 Systems is ideal chatbot testing ground given standardized question patterns - Chatbot hesitation due to risk of angering already frustrated customers **2. Living Documentation & RAG Bot** Support team currently has to go to the website, find documentation, and pass it to people for internal use. Need a centralized, always-current documentation source. - Sales team currently reaches out to support to ask product questions - A bot with all documentation/RAG at its fingertips would let sales self-serve - FAQ section is outdated and was ignored during website relaunch - Michiko (external contractor) managing technical website rebuild including new FAQ area - Documentation is overly technical — could benefit from conversational tone matching email responses - No single place where all critical information lives and is easily accessible **3. Product Support Request Tagging (NetSuite) — Replacement Orders? ** When replacement parts need to be sent, the data captured in NetSuite is insufficient for pattern detection. Need signal-based reporting. - NetSuite replacement orders have only one dropdown field with ~15 vague options - Cannot record multiple issues per replacement order (e.g., "4 LEDs broken + shipping damage") - Free text descriptions don't align with actual categorization - Currently takes days of manual email searching to identify patterns (e.g., discovering breakage was color-specific) - Desired state: Telemetry-style reporting — "This batch for this product in this color has had 4 replacement orders in 3 months → flag it" - That feedback loop should tie directly back to manufacturing/production - NetSuite consultants meeting overdue by ~2 months — Molly has documentation prepared on desired reporting structure **4. Product Development Workflow & Change Tracking** Reinforces the #1 value-add identified across interviews. The product lifecycle lacks formal tracking and sign-offs. Current flow: Omer + Jay (Ideation, secretive phase) → Brian's Pro Dev team (Productionizing) → Iterative documentation → Multiple departments Who needs to receive info from Pro Dev: - Mandy (Designer) — builds spec sheets, currently guessing due to incomplete info - Molly (Support) — needs to know what changed to support customers - Julienne (Compliance) — handles all compliance/certification status - Design team for documentation and approvals Key problems: - No digital sign-off system — designer can get Randy's approval without notifying Brian (PM) - Changes to existing products go unrecorded - Part changes affecting 20+ products are typically missed - Randy and Omer make ad-hoc decisions in random conversations without documentation - Randy sometimes forgets decisions - Departments operate as islands with limited cross-team visibility Monday board concept discussed: - Single object tracks product from ideation through production - Multi-approver sign-offs at checkpoints - Tied to Slack notifications for all relevant parties - Key implementers: Brian (Pro Dev), Molly (documentation), Tyler (operations) - Challenge: team perception that added process slows them down **Company Structure Detail** **Two Business Lines Under One Organization** Both funnel through NetSuite for core operations. **Tech Stack Notes from Interview** - Hiver: Gmail overlay for support — custom fields, AI/automation options available but underutilized. Previous Zendesk attempt created more problems. - Shopify: Powers 22 Systems e-commerce only - 3D Experience/PDM: Poor search functionality, team divided on usage, must know exact product name to find info - CRM opportunity: Layer over Gmail/Hiver to enable BI reporting while maintaining personal support brand. Flow: Hiver/Gmail → CRM → NetSuite → Looker Studio dashboards - Monday.com: Mentioned as concept for product workflow tracker **Recommended Next Conversations (from Molly)** - Bryan (Brian Magennis) — Lead product developer/project manager - Daniel (Nikitiuk) — Designer, interested in systems, formerly systems consultant - Mandy (Chang) — Keeper of all documentation - Sachi (Sachie Sakai) — Keeper of website materials for Bocci side - Julian — Certification tracking, phones, support (on vacation starting Monday) - Connor (NetSuite admin) — NetSuite administrator, schedule consultants meeting - Andrew (Nguyen) — Production/manufacturing **Action Items** This interview strongly reinforces the #1 Value-Add (Change Management Process) and adds concrete detail about the product development workflow, support pain points, and NetSuite limitations that should feed into Phase 1 recommendations.

Interviewee: Apollo — OAO / Omer Arbel Office (Architectural Designer) Date: 2026-02-25 | Duration: ~44 min | Format: In-person Reports to: Omer Arbel | Email: apollo@omerarbel.com | Tenure: ~5–6 years **TL;DR** Apollo was the window into OAO — the creative/architectural practice sitting directly under Omer. Pre-interview agenda was building science (solar heat gain, sunpaths, code). What actually emerged: OAO's core daily problem is the same as everyone else's — communication breakdown and decision tracking. Apollo maintains an 18-page meeting minutes document that nobody reads. He wants AI to let contractors query it directly. The building code + rendering insights are real quick wins. Omer has asked Connor to wait on OAO-specific work until Bocci-wide is further along. **Key Findings** **1. Communication, Not Building Science, Is the Priority** Apollo immediately redirected: "Those are like not really important for the day-to-day stuff. I think the priority list is like communication." Named two structural problems: cross-team communication fragmentation, and the Omer/Randy decision-making dynamic where two principals may not be perfectly aligned and the team has to translate both. **2. Omer/Randy Tension Is Structural** Omer sets the design goal. Randy is both client and boss. Two decision-making parties that may not align — and the team has to navigate both. Decisions get made verbally; harder unresolved items get dropped for low-hanging fruit. "Tracking those conflicts and keeping record of them would be nice." **3. Day-to-Day Is Fire-Fighting** Apollo spends most of his day piecing together a puzzle of decisions from meetings he wasn't in. Avoids formal meetings (they get billed to Governor's Point as contractor hours). Defaults to texting individuals for quick answers. Nothing gets logged. Communication bottleneck is internal, not external. **4. The 18-Page Meeting Minutes Document Nobody Reads** Apollo maintains a rolling Google Doc of all project decisions. Highlights new items in red. 18 pages. Nobody reads it — "It's seen as busy work." He wants people to be able to query it and get instant answers: "If the guys had questions down in Bellingham, they could ask AI, 'What was decided on this?' And it could find it." He's been manually building a decision-log RAG system for years without the infrastructure. **5. OAO Practice Structure + Pipeline** Partner-based, not client-based. Omer decides what projects align with the practice vision. 5–6 active projects, each siloed with one lead. Governor's Point: 5 years design + 3 years construction = decade-long project. Apollo sees efficiency opportunities: "I think we can truncate it. There's some efficiencies there." Pipeline: Omer concept → Jaden early physical model (proves buildability) → OAO architects produce drawings → contractor mock-ups. **6. Tools: Revit + Rhino + Early ACC Adoption** - Revit: Parametric drafting. Contractors need 2D floor plans/elevations. - Rhino: 3D ideation and complex geometry. One drawing change = 20 2D drawings to update manually. - ACC (Autodesk Construction Cloud): Introduced ~1 year ago. Has version tracking, asset management, meeting minutes. Slow adoption. "It'd be nice to have one place for everything." - Google Drive: Home for the rolling meeting minutes doc. **7. Rendering Takes Days** One view of one building elevation: Rhino → render plugin → Photoshop layer adjustment = days. Apollo wants AI sketch-to-render for quick internal ideation. Not for publishing — just for "taking existing screenshot rendering from Rhino and making it look slightly better." **8. Building Code and AI** OAO works extensively with building code (VBBL + provincial + AHJ-specific). Apollo's insight: code tells you what you can do, not what you can't. You have to infer the intent. "The rule is less important than the reason for the rule is how we approach it." Wants AI that gives code answers + citations + intent rationale. Also wants AI cost estimation (Yardsticks document) for quick per-square-foot estimates before committing to expensive expert consultations. **Quick Wins for OAO** - AI-assisted decision log — structured capture of Omer/Randy decisions with search; replaces the 18-page Google Doc - AI sketch-to-render — quick internal visualizations without the full render pipeline - AI building code query tool — answer + citation + intent - ACC adoption — tool already licensed; barrier is behavioral **Scope Note** Omer has asked Connor to wait on OAO-specific work until Bocci-wide engagement is further along. Apollo is aware and open to cross-applying Bocci learnings. **Key Quotes**

**Jasmine Whelan — Product Development (Bocci Dedicated)** Date: Feb 18, 2026 | Duration: ~45 min Reports to: Brian Magennis | Location: Vancouver Office Interviewer: Connor Pundick | Also present: Daniel **1. Role & Context** - Typical week: Hardware design — sketching concepts in InDesign, 3D modeling in SolidWorks, 3D printing prototypes, iterating with Omer on aesthetics and Paul on assembly/functionality - How long at Bocci: ~3 years (approaching 4). Newest member of the design team. Started at 23-24. - Works most closely with: Omer (creative direction), Paul (functional testing), Lana (coordination/materials), glass shop team, China manufacturing team. Day-to-day with Les, Daniel, and Julien within ProDev. - What they enjoy most: Creative problem-solving and design breakthroughs. The moment when a new mechanism or approach clicks. - Notable: Only person doing 100% hardware design work on the Bocci side. Not assigned to 22 Systems projects but provides input on some aesthetic decisions. Connor's read: Jasmine is junior, eager to please, and genuinely excited about this engagement. She rambled during the interview — not evasive, just a creative thinker who processes out loud. May not be the most organized person; could benefit from structured workflows and guardrails she doesn't have to think about. Her brain works in creative bursts rather than checklists. Notes: Jamie AI misidentified Jasmine as "Daniel" throughout the transcript. All references to "Daniel" in the source transcript refer to Jasmine Whelan. **2. Communication & Decisions** - Information flows informally — "not a formal set process, communication happens organically based on who needs to know" - Design changes trigger cascade across multiple departments: Jasmine → Omer (aesthetics) + Paul (functionality) → Lana (materials/ordering) → Tyler (production), Mandy (instructions), suppliers - Brian/Ryan critical for managing timelines and communicating priorities between ProDev and leadership (Randy, Omer) - Design meetings historically every 2 weeks with Omer, Paul, Lana, glass shop, production — but paused 2+ months for Milan push. Expected to resume after showcase. - China manufacturing: daily communication. Language barrier significant — in-person/calls more effective than email. Thomas (QC in China, speaks Mandarin) serves as translator. Brian and Patty visit annually. Connor's read: Jasmine was notably unbothered by the informal information passing — likely because she's a contributor to the informality rather than a victim of it. She doesn't feel the pain of gaps because she's not on the receiving end. This contrasts with what Tyler, Mandy, and Molly have said about information getting lost. Notes: Jamie misidentified Brian Magennis as "Ryan" in the transcript. There is no Ryan — all references to "Ryan (Product Development Manager)" are Brian. **3. Tools & Systems** - Tools used daily: SolidWorks (primary 3D CAD), InDesign (2D concept sketches), ChatGPT (30-50 prompts/day) - Tools supposed to use but don't: 3D Experience (PDM system) — adopted ~3 years ago after a near-data-loss incident but team has largely rejected it - Workarounds built yourself: Saves all CAD files locally to shared server folders during active development. Only pushes to 3DX when designs are finalized for production (~once every 2 months). This is the team-wide workaround, not just Jasmine. - Magic wand fix: AutoCAD design translation / AI image generation - AI tool usage: ChatGPT power user — 30-50 prompts/day across work and personal. Work uses: simplifying technical emails to China, material specs/tensile strengths, learning professional email writing. Currently on free tier (limited prompts, one image upload per 12 hours). Company exploring business accounts. Tried Midjourney (lacked design thoughtfulness), Gemini, Claude, and Grok (once each, not explored further). Connor's read: The ChatGPT attachment was the surprise of this interview. Jasmine isn't just dabbling — she's a genuine power user, rare in a manufacturing org. She wants ChatGPT for the whole company. This is a strong signal for AI adoption readiness on the ProDev team. Notes: Aaron Miller (recommended as next interview) apparently introduced ChatGPT to the team and is an early adopter. He could be a key champion for AI rollout. **4. Workflows & Pain Points** **The 141 Product (Case Study in ProDev Lifecycle)** - Complete hardware redesign — invisible tapered suspension + magnetic pogo-pin LED mount - 2-year development cycle, launched Milan April 2025 - Now in T1 (full production) but back-ordered — unexpected demand + rejected glass batch created supply shortage - Required complete redesign shortly before launch due to assembly finickiness + Omer's aesthetic requests - Extensions in development: sconce version for Milan, tree versions being tested - Glass shortage forcing prioritization decisions on which extensions make it to Milan **Product Development Flow** - Omer initiates with directional input → Jasmine sketches/InDesign → SolidWorks 3D modeling → 3D prints for physical evaluation → iteration cycles (weeks, in-person meetings) - 3D printing: in-house for initial concepts, external companies for higher-quality nylon prints (3-day lead time, sometimes next-day rush) - Project volume: 3-4 high-priority projects in busy months, ~4 low-priority in slow periods, ~8-10 major projects/year - Timeline pressure: Milan announcements force rushed launches, products get high-priority status immediately after public exposure **3D Experience Pain (Deep)** - Crashes frequently, requires saves every 10 minutes - Saves take up to 2 minutes, breaking workflow momentum - File referencing issues — doesn't always pull correct file versions - China manufacturing and production teams rejected it immediately as too complicated - Team attempted daily use initially but gradually pushed back due to productivity loss - Team knows migrating away would be costly but is open to change - Decision rests with Brian, Randy, and Omer **Information Flow Gaps** - No centralized location for design ownership — spreads through word-of-mouth - Mondays (PM tool) status unclear — thinks Brian manages it now - Design research relies on Pinterest (time-consuming, only Jasmine uses it, team misses trends) **5. 3D Experience & File Management** - 3DX brought in ~3 years ago after near-data-loss incident. Intended as cloud PDM. - Reality: team tried daily use, rejected it. System is unintuitive, crashes, slow saves, broken file referencing. - Current workflow: save locally to shared server CAD folders during active development. Push to 3DX only at production sign-off (~once every 2 months). - Jasmine's position: she detests it. Would welcome a replacement. - Migration concern: team aware it would be costly/time-consuming but open to it. **6. Aspirations & Closing** - What should NOT change: Company culture — collaborative, easygoing, low politics. People can approach colleagues openly. - If she could fix one thing: Not explicitly stated, but the strongest energy was around AI tools (ChatGPT access/business account) and replacing 3D Experience. - Anything nervous about: Not expressed. - Key recommendation: Interview Aaron Miller next — deep AI/tech knowledge, early adopter, expect 1.5-2 hour session. **Post-Interview Capture** - What did Jasmine care about most? AI tools (ChatGPT specifically) and getting rid of 3D Experience. She wants better tools and she's already proved she'll adopt them. - Top 3 pain points: - Actual tools used: SolidWorks, InDesign, ChatGPT (30-50x/day), Pinterest, 3D printers (in-house + external), 3DX (reluctantly, ~bimonthly) - Change readiness: Champion — actively using AI, excited about change, wants ChatGPT for the company - Quick wins: - Surprises: Depth of ChatGPT usage (30-50 prompts/day). Junior designer is the most active AI user encountered so far. - Confirms or contradicts prior interviews: - Key quotes: - New stakeholders identified: Source transcript: Jasmine Welan Interview (Jamie) — Note: Jamie misidentified Jasmine as "Daniel" throughout

Interviewee: Miya Kondo — Sales Director, 22 System Date: February 25, 2026 Interviewer: Connor Pundick Source: Jamie transcript + Connor debrief **Role & Context** Miya is the Sales Director for 22 System, joined around 2020. She manages the full 22 System sales function — ~70% direct-to-consumer via Shopify, plus 10-15 dealer POs entered manually into NetSuite daily. She previously also covered Bocci dealers and OEO Works (Japan), and handed off the Japan territory to Byron around early 2025. She manages multiple inboxes: mia@bocci.com, info@22system.com, 22system-support, and 22system-eu. Connor's read: Open, warm, a perfect culture fit. She is content and happy where she is — a clear example of what Connor is calling the "cult of Bocci" pattern: people who have been here long enough to normalize the chaos, stay for the people and the culture, and don't fully connect the dots between the warmth they feel and the structural dysfunction creating their pain. **Key Findings** **1. Product Spec Changes Happen Without Sales Visibility** The most concrete example in the interview: the on/off switch product was communicated to Miya as available in both 35mm and 46mm diameters. She later discovered only 46mm was being made. She escalated to Randy and received verbal and email confirmation that 35mm would be added — then, without her knowledge, the decision reverted to 46mm only. Root cause: side conversations between product development and Randy change decisions with no read-out to affected teams. A cross-functional meeting was recently convened (Tyler, Julian, Miya, Victoria, Erin, Lana) specifically to improve alignment. Randy didn't attend. Connor's read: This isn't just a communication gap — Randy's absence from the alignment meeting is telling. The problem is structural: decisions flow through Randy informally and can be changed by him at any point, with no obligation to notify downstream teams. **2. No CRM — ~$15M Portfolio Managed Manually by One Person** No formal system for tracking active sales deals or opportunities. When a new inquiry arrives, Miya manually searches email history, then checks NetSuite or Shopify separately. Past Bocci orders live in NetSuite; 22 System orders live in Shopify — no consolidated view at the dealer level. Dealer discount structure is inconsistent and not formula-based. This is a clean confirmation of Theme 2 from the strategy master. **3. Trade Account Management Entirely Manual** Miya identifies trade customers by email domain or business references, manually creates individual discount codes, and uses a PDF registration form as a friction filter for ambiguous cases. A previous generic code reached 10,000+ uses with zero tracking before she replaced it. No automated portal or database tracking exists. **4. Information Silos Create Constant Interruptions** Miya has no visibility into: incoming inventory from China (must ask Lana), product development status or ETAs (must email Julian), and upcoming release timelines. Several months can pass before she realizes a product's status has changed. 50+ emails/day across 4 inboxes. Connor's read: The information flow problem isn't unique to Miya, but her case illustrates the cost at the sales layer — she can't proactively sell or prepare if she doesn't know what's coming. **5. Marketing Collaborations: No Agreements, No Recourse** Miya tracks marketing collabs (free products or large discounts for cross-promotion) in a folder. Major problem: deliverables often not received, delayed up to 2 years. She attempted a formal marketing agreement contract — partners pushed back, and formal agreements create tension. Currently no recourse when partners don't deliver. **6. 22 System Is Operationally Simpler Than Bocci** The purchasing workflow for 22 System is significantly simpler than Bocci. Shopify-NetSuite auto-sync handles ~35 orders/day well. No major process standouts on the 22 System side. Connor's read: Job creep is evident — Miya manages tasks well beyond what's likely in her JD, including support inboxes and multi-brand coordination. But per Connor, this is standard for the org. **7. Culture and People: The "Cult of Bocci"** Miya explicitly cited staying "for the people" as a primary reason for tenure. She praised low turnover, minimal red tape, and organic communication — contrasting it favorably with her prior experience in fast fashion (COS/H&M). In the same breath, her biggest pain was the lack of a product roadmap and cross-department communication. Connor's read: She is not connecting these dots. The warm relationships are real, but they're also the mechanism by which systemic dysfunction gets tolerated. The founders' desire for absolute control creates last-minute changes and organizational chaos; people like Miya absorb that friction personally rather than naming it as a structural problem. This is a broader pattern across the org — Connor is calling it the "cult of Bocci." **8. AI Usage: Claude User** Uses Claude for copywriting and newsletter copy variations. Prefers Claude over ChatGPT/OpenAI ("OpenAI/ChatGPT are evil, Claude seems more benevolent"). Learned about Claude from Julien (Minga) a year or two ago. Overall attitude: pragmatic, slightly cautious, open to beneficial tools. **Themes Confirmed** - Theme 1 (Siloed Information): Confirmed — no visibility into ProDev status, inventory ETAs, or cross-brand customer data. - Theme 2 (Manual Sales & No CRM): Confirmed — single salesperson managing $15M portfolio via email and spreadsheets. - Theme 3 (Change Ledger): Confirmed — switch diameter reversal is the cleanest example yet of an undocumented Randy decision cascading into sales impact. **Change Readiness** 🟡 Neutral — Content and happy, not pushing for change, but not resistant. Aware of the pain points but accepts them as normal. Not a champion or blocker. **Action Items** None. **Open Questions** - Is the recent cross-functional alignment meeting (Tyler, Julian, Miya, Victoria, Erin, Lana) ongoing? What came out of it? - Is Randy's non-attendance at that meeting a known pattern or a one-off?

**Daniel Nikitiuk — Design Lead, 22 System** Date: Feb 18, 2026 | Duration: Reports to: Brian Magennis | Location: Vancouver Office Context: 22 System design lead and 3D Experience champion per Brian. Formerly a systems consultant before becoming a designer — potential strong ally for rolling out process/systems changes. Per Molly: interested in systems. Tuesday mornings: Daniel + Julien meeting on 22 System strategy. **1. Role & Context** - Typical week: Design work for 22 System products, CAD administration, PLM administration, solution architecture. Multiple hats. - How long at Bocci: Since ~2022. Formerly a systems consultant before becoming a designer. - Works most closely with: Brian Magennis (reports to), Vince, Julian (22 System team). Tuesday morning strategy meetings with Julien Mainguy. - What they enjoy most: Systems and process design. Sees himself as the PLM/governance champion. Notes: Daniel holds multiple roles: lead designer for 22 Systems, CAD administrator, PLM administrator, solution architect. The multiple hats create delivery effectiveness issues by his own admission. Came prepared with 3 formal documents making the case for a "Director PLM / Design Systems Lead" role. **2. Communication & Decisions** - 22 System has structured approach using 3D Experience change actions with formal approval routes and maturity states. Julian and Brian review and sign off electronically. 60 change actions released last year. - Bocci operates ad hoc — files sent directly to China without tracking. No formal revision history. 93 may be first Bocci product to adopt structured process. - Document flow to Mandy is always rushed because production is beginning and sales wants speed. Requires rework when designers make changes. - Julian handles documentation and sales alignment timing. Certification typically marks the point where design changes stop. Notes: The 22 System process is genuinely more mature — 60 formal change actions is real output. The gap is that Daniel frames the Bocci side’s failure to adopt as a governance/authority problem rather than a tool/culture fit problem. He blames other teams for not following the process. **3. Tools & Systems** - Tools used daily: SolidWorks within 3D Experience, G Suite, Adobe Suite, Scribe (SOPs — just Daniel), Notion (just Daniel for tracking tickets), ChatGPT (personal account, requested pro dev account but stalled) - Tools supposed to use but don't: N/A — Daniel uses everything; the issue is others don't. - Workarounds built yourself: Created online training series, video tutorials, SOPs, one-on-one mentorship for 3DX. Knowledge lost through inactive use. - Magic wand fix: Enterprise AI policy. Authority to enforce PLM adoption. - AI tool usage: ChatGPT for research. AI being built into SolidWorks platform as next development phase. Notes: Daniel has put real effort into training materials and SOPs. The problem isn't that he hasn't tried — it's that the training doesn't stick because people don't use the tool consistently enough to retain it. This is a classic adoption death spiral. **4. 3D Experience Deep Dive** - Two parallel release processes running: 22 System (structured, 3DX) and Bocci (ad hoc, no tracking). - Governance is the primary barrier per Daniel — no enforced authority, designers can choose ad hoc vs structured without consequences. - Metadata and part number assignment is critical but failing — when not assigned at part level, they don't appear on drawings, breaking traceability. - File management: current Bocci problem is files released completely outside the system. New products not entered into platform. Daniel doesn't know where Bocci files are physically stored. - Manufacturing files exported as STEP + PDF to China. Initially tried direct 3DX access but Chinese government internet throttling caused latency. SD-WAN identified as future solution. - Save process friction: all team members work locally first, then manually save to cloud. 30 sec to 1 min upload. Daniel doesn't believe this justifies avoiding the platform. Notes: Daniel's diagnosis of the file management chaos is accurate — Bocci files living outside the system creates the exact "institutional amnesia" and mismatch problem the PLM was meant to solve. But his solution (enforce compliance through authority) ignores WHY people aren't using it. Jasmine said the team "rejected" 3DX. Mandy called it "the worst thing ever." Brian said 2-3 of 7 use it. The tool has a user experience problem, not just a governance problem. **5. Systems Thinking & Process** - Daniel's #1 priority: governance and adoption of 3DX. Assign formal responsibility for PLM. - His #2: inability to have authority in the process. Needs either a separate role with authority or explicit authority to enforce. - Brought prepared documents: (1) "Why Does Change Fail" analysis, (2) "Responsibilities for Director PLM / Design Systems Lead" job description, (3) Executive Summary arguing adoption & governance are the primary capability gap. - 22 System process is transferable in principle but requires enforcement that Bocci's culture resists. - Manufacturing timeline ballooned from 6 to 8-16 months due to added verification gates. Quality checklist exists but only ~25% completed before release. - No formal project management across either product line. No timelines, deliverables, or date accountability. - Historical issue: China made unauthorized tooling changes without notification, causing production mismatches. Daniel's structured process with China team has actually worked well to address this. Notes: The China manufacturing control story is where Daniel's approach has genuinely delivered value. Structured injection molding process with stage gates and money release points. The unauthorized tooling change incident (barrel alignment causing 10-15 degree misalignment) was caught and the person responsible was fired. This is a real win for process. The problem is Daniel wants to apply the same enforcement model internally, and Bocci's culture won't accept it. **6. Aspirations & Closing** - Wants formal PLM/Design Systems Lead authority. - Sees himself as the solution to the governance gap. - Willing to share release process documentation and quality checklist files. - Follow-up meeting scheduled. Notes: No explicit mention of what should NOT change. Revealing — Daniel frames everything through what needs to change (with him in charge), not what to preserve. **Post-Interview Capture** - What did Daniel care about most? Getting formal authority over PLM governance and 3DX adoption. He wants a "Director PLM / Design Systems Lead" role with enforcement power. - Top 3 pain points: - Actual tools used: SolidWorks/3D Experience, G Suite, Adobe Suite, Scribe (SOPs), Notion (personal), ChatGPT (personal) - Change readiness: Champion (for his specific vision) / Resistant (to alternative approaches). Daniel is deeply invested in 3DX as THE solution and positions himself as the authority. He is not open to considering that the tool itself may be part of the problem. - Quick wins: Quality checklist enforcement (exists but not completed), 93 Bocci as pilot for structured release process - Surprises: Brought 3 prepared documents — executive summary, job description, and change management analysis. This was a pitch, not an interview. Manufacturing timelines 6→8-16 months is a significant data point. The China manufacturing control process (stage gates + money release points) is a genuine success story. UL certification requires full factory assembly — can't do final assembly outside certified facility. - Confirms or contradicts prior interviews: - Key quotes: (from Jamie transcript) - Connor's read: Intense conversation. Daniel is polarizing — a fantastic IC who struggles with people. On numerous occasions said he simply needs the power to tell people what to do and things will be better. Blamed other teams for not using the tool properly and wasn't open to another worldview. Danced around the feedback that others have struggled. Everything he said contradicted other teams. His approach felt very much like "everyone else needs to change for me." There IS truth in his diagnosis — 22 System genuinely has a superior release process. But his prescription (give me authority to enforce) is antithetical to Randy and Omer's culture of creativity, flow state, and fluidity. Randy came in after the meeting and said nothing notable about Daniel, but immediately pointed Connor to Andrew Nguyen as someone worth listening to. Leadership doesn't endorse Daniel's approach. Source transcript: Daniel Nikitiuk — Design Lead, 22 System

Interviewee: Michiko — Web Contractor (bocci.com + omararbel.com) Date: February 24, 2026 Interviewer: Connor Pundick Source: Jamie transcript + Connor debrief Note: Jamie labeled Michiko as "Speaker 5" throughout the transcript. Also rendered "Claude Code" as "Cloud Code" and "Bocci" as "Bacci" in several places. **Role & Context** Michiko is an external contractor managing bocci.com and omararbel.com on a 20-40 hrs/month maintenance contract. She was originally brought in for the website rebuild and stayed on to manage ongoing maintenance and new feature development. She works daily with Sachi and Mandy via email and Slack. Connor's read: Smart, nice, smooth sailing. She has what she needs and does her job well. The website is a relatively stable part of the operation compared to the dysfunction elsewhere in the org. Teams that interface with the website can generally do what they need without friction. **Key Findings** **1. The Configurator App Is the Most Technically Fragile Component** The configurator is a React app with a Craft CMS backend integrating Shapediver for 3D visualization. It allows professional designers to create consultation PDFs by tinkering with how constellations of lights can be laid out. It's stable when left untouched but fragile when changes are needed — requires coordination between Gustavs (Latvia, React dev), Albert/Alp (Switzerland, 3D GLTF modeling), and the Shapediver team. 3D modeling work is time-intensive and has delayed launch. Current 3D model management via JSON files and Google Sheets is inefficient; workflow improvements planned. Connor's read: This is an interesting and valuable product surface. The fragility is a contractor coordination risk, not a people problem. **2. NetSuite Integration for Bocci Products Not Yet Live** 22 System is connected to NetSuite on the website side. Bocci (28/20 series) is not — trillion+ SKU variations make automation complex. Ivars (Latvia) has built the integration tool but it hasn't launched yet. Current focus is on integrating portable sales and parts first. This is relevant context for the broader Craft → NetSuite integration work Conor Graham has flagged. **3. Content Workflows Run Through Google Sheets — Working Well** Product management flows primarily through Google Sheet imports rather than direct CMS editing. Sachi and Mandy prefer the collaborative sheet workflow for reviewing changes. Multiple import flows cover Lighting Products, NetSuite Inventory Items, Configurator/Color App data, and FAQ/Resources. Dynamic pricing calculates based on selected options, occasionally requires debugging. FAQ + Support consolidation is launching in March — migrating to a sheet-based import combining both into a single unified page. Main blocker was front-end layout alignment, finalized on the day of this interview. **4. Contractor Ecosystem** - Gustavs (Latvia) — React developer, configurator and color apps - Ivars (Latvia) — NetSuite integration API work - Albert/Alp (Switzerland) — 3D GLTF pendant files for configurator - Civilization (distributed) — UX/UI consultation - Burn Kit (Vancouver) — manages 22 System Shopify site (90-95% of development); Michiko assists with Klaviyo integration and multi-language work **5. Upcoming Launches** - Series 93 — launching March - New Color app — launching post-Milan trade show; old Color app deprecated after new version goes live - FAQ/Support consolidated page — launching March **6. DAM Redundancy — Minor Flag** Data redundancy exists across Digital Ocean Spaces, Swivel server, and local servers. Michiko noted this as a consolidation opportunity for a more uniform workflow. Aligns with the broader DAM modernization initiative already on the roadmap. **7. Uses Claude Code** Michiko has recently adopted Claude Code for automating routine tasks — image tag updates, file deduplication. Jamie's transcript rendered this as "Cloud Code" — confirmed as Claude Code from context. She expressed curiosity about further AI integration possibilities. Team enjoys the Google Sheets workflow; improvements should maintain that preference while adding automation. **Themes Confirmed** - Theme 1 (Siloed Information): Partial — DAM redundancy flagged, but the website content workflow is actually one of the more functional information flows in the org (Google Sheets as source of truth). - Theme 2 (Manual Processes): Light confirmation — some manual debugging of dynamic pricing, but overall website operations are more automated than most departments. - Theme 3 (Change Ledger): Not surfaced — website changes go through a proper pull request workflow between Michiko and Gustavs. **Change Readiness** 🔵 Open — Curious about AI integration, receptive to improvements, already using Claude Code. Not a champion pushing for org-wide change, but a pragmatic adopter. **Action Items** - Michiko to send Connor: Burn Kit email contact, and access to Google Sheets (Pricing, Product, FAQ, Resources) - Connor believes these have been received. **Open Questions** - When does Ivars' NetSuite integration for Bocci products go live? Is there a timeline? - Is the configurator app publicly available, or designer-only? - What is Michiko's last name? (Not captured in transcript or Jamie notes.)

**Interview: Maxwell Bishop — Digital Advertising, Socialistic** Date: March 2, 2026 | Interviewer: Connor Pundick Role: External — runs Socialistic (performance marketing agency). Bocci is one of ~15-20 clients. Relationship: Close business partner of Connor's. Conversation was fully frank — no diplomatic filtering. **Key Findings** **1. The Attribution Gap Is Total** ~12,000 monthly site visits. 0.3% online conversion rate. But the sales team (Emily + Sachie) receive ~50 emails/day through Hiver — and there is zero tracking connecting whether those emails are marketing-generated. The clearest example: a $20,000 Pinterest-attributed order had already been in the sales pipeline via email for three weeks. Attribution couldn't determine whether the ad drove the email or whether the email was already in flight independently. Connor's read: This surprised him. Max doesn't know if any conversion that flows through Hiver has a digital ad upstream. The funnel goes: ad → website → email to Hiver → manual follow-up. The middle section is completely dark. This confirms Theme C not just as a feeling but as a structural measurement failure. **2. Net-New Only Strategy Has a Structural Blind Spot** All paid campaigns exclusively target net-new customers. Existing customers, recent site visitors, brand followers — all excluded. Philosophy: "if you love a brand, you'll come back without being paid to." Consequences: - No B2B lead nurturing through digital - No remarketing to cart abandoners (email should handle this, but Klaviyo isn't set up for it) - No tracking of customer lifecycle after first purchase — nobody knows if 14P buyers eventually upgrade to chandeliers **3. Klaviyo Is an Untapped Asset** 48,000 total profiles. 40,000 active and marketable. 2,400 confirmed customers (at least one purchase). Sources: website signups, VIP events, warehouse sale (April 2023), Milan Aperitivo. The abandoned cart flow for 14P currently sends a design consultation CTA — the right message for a $100,000 chandelier, completely wrong for an impulse product. Max is working to fix this. Proposed 3-email sequence for 14P abandoned cart: - Simple reminder — "You left something in your cart" - Support offer — "Can we help?" - Craftsmanship story — making-of video, behind-the-scenes process Email economics: ~$0.50/email vs. full paid retargeting cost. Bocci's existing storytelling assets are perfectly suited to this. **4. Budget Scaling Blocked by Data Blindness** Max is hesitant to recommend significant budget increases because without attribution data, there's no way to know where they sit on the diminishing returns curve. His framing: "If diminishing returns happen at 10 and we're at 9, spending another $50k/month could push us past optimal." Without data, the upside is unknowable. **5. HubSpot as the Missing Link** Meeting with HubSpot this week (Connor + Max). Goal: - Backfill historical sales/support emails from Hiver into contact records - Implement contact forms with UTM capture to replace mailto: links (current mailto: setup breaks UTM tracking entirely — when someone clicks support email, all campaign attribution is lost) - Connect marketing-generated traffic to actual conversions for the first time **6. The Louis Vuitton Playbook (Untapped)** Max's long-standing pitch: advertise aspirational premium pieces (chandeliers) to build brand desire at the top of funnel, then convert with accessible products like 14P. LV markets runway collections, sells wallets. Bocci markets 14P almost exclusively. Premium advertising is largely untapped because there's no CRM infrastructure to manage the longer sales cycle. **7. B2B Blind Spot** Digital advertising is 100% consumer-focused. No mechanism to generate or nurture B2B leads (dealers, design consultants) through digital channels. Charged design consultation idea ($150, refunded on purchase or returned if not) has been pitched for years. Would require CRM to work. **8. Reporting Stack in Transition** Currently using Funnel to aggregate platform data. Moving away — cost without proportional benefit. Evaluating Looker Studio. Connor demo'd the Cloudflare portal, which landed well. Max already thinks in data ontology terms — the lakehouse framing resonated immediately. **9. Randy Relationship & Adoption Dynamics** Randy is Max's primary internal contact. He's casual, trusting, doesn't micromanage metrics. Max confirmed the slow-but-then-fully-in adoption pattern at Bocci: historically took time to buy into digital advertising, now fully committed. Connor's read: This is historical context, not a current blocker. Randy is already bought in on AI and systems improvement for this engagement. The pattern is worth noting for framing future recommendations but doesn't change the current approach. **Theme Confirmations** - Theme C (Flying Blind on Growth): Fully confirmed and quantified from the external seat. The attribution gap is structural, not incidental. - Theme 2 (No Unified Customer View): Confirmed from a new angle — the marketing layer sees what happens before a sale but has zero visibility into what happens after. The customer journey has no data at any stage. **Action Items** **Change Readiness** 🟢 Champion — external, but arguably the most data-forward voice in the entire engagement. Has been pushing for CRM, attribution infrastructure, and AI-assisted analysis for years. Potential future business partner with Connor.