How I AI

How a VC and tech founder used AI to launch a brick-and-mortar business in their spare time

Andrew Mason (founder of Groupon, now CEO of Descript) and Nabeel Hyatt (General Partner at Spark Capital) teamed up to open a physical board-game social club in Berkeley, with AI as their business partner. In this episode, they break down how they used Claude to generate a full business plan, model financials, plan the space layout, navigate Berkeley permitting, categorize hundreds of games using a custom Dewey Decimal–style system, and build an AI concierge that matches players with games via text. They also share how working on this side project helped rewire how they use AI in their day jobs—and why more people should use AI to build real-world things. *What you’ll learn:* 1. How to use Claude Projects as your business copilot to create comprehensive business plans, financial projections, and space layouts 2. A workflow for categorizing hundreds of board games using an AI-generated “Dewey Decimal System” that makes game discovery intuitive 3. How they built an AI concierge service that matches players with games and coordinates group play sessions via text message 4. Why AI enables side projects that would otherwise be impossible due to time constraints and specialized knowledge requirements 5. A simple system for creating customer personas that inform your business model and event programming 6. How to use model context protocols (MCPs) to connect AI assistants to business tools like Airtable without complex coding *Brought to you by:* Lovable—Build apps by simply chatting with AI: https://lovable.dev/ Persona—Trusted identity verification for any use case: https://lovable.dev/ *Where to find Andrew Mason:* LinkedIn: https://www.linkedin.com/in/andrewmason/ X: https://x.com/andrewmason *Where to find Nabeel Hyatt:* LinkedIn: https://www.linkedin.com/in/nabeelhyatt/ X: https://x.com/nabeel *Where to find Claire Vo:* ChatPRD: https://www.chatprd.ai/ Website: https://clairevo.com/ LinkedIn: https://www.linkedin.com/in/clairevo/ X: https://x.com/clairevo *In this episode, we cover:* (00:00) Introduction to the board-game social club concept (02:44) How AI made a challenging side project possible (06:14) Using Claude as a business copilot for planning (12:53) Developing customer personas with AI (15:45) Using AI to determine business viability (21:02) Navigating Berkeley real estate and permitting (25:18) Building an AI concierge for game matchmaking (28:10) Database design with Airtable for non-technical founders (32:04) Creating a custom board-game categorization system (36:20) Demo of the text-based AI concierge service (40:38) Enabling experiences that wouldn’t exist without AI (43:42) Lightning round and final thoughts *Tools referenced:* • Claude: https://claude.ai/ • Airtable: https://airtable.com/ • n8n: https://n8n.io/ • Twilio: https://www.twilio.com/ • Cursor: https://cursor.sh/ • Windsurf: https://www.windsurf.io/ • Python: https://www.python.org/ *Other references:* • Model context protocol (MCP): https://www.anthropic.com/news/model-context-protocol • Tabletop Library: https://tabletoplibrary.com/ • Descript: https://www.descript.com/ _Production and marketing by https://penname.co/._ _For inquiries about sponsoring the podcast, email jordan@penname.co._

AMAndrew MasonguestNHNabeel HyattguestCVClaire Vohost

Aug 4, 2025·48m·Watch on YouTube ↗

📖 CHAPTERS

1. 1
   0:00 – 2:44

   Board-game renaissance and the Tabletop Library social club idea

   Andrew and Nabeel describe how modern board games became a compelling social outlet and why traditional game nights were hard to organize. They land on a membership-based, physical “show up and play” club concept that also helps members find others to play with.

2. 2
   2:44 – 6:14

   Why AI made the side-project retail business feasible at all

   They argue the business would not exist without AI because neither founder has time or prior small-business experience. AI (especially Claude) provides momentum, structure, and rapid feedback cycles that make a “pull the thread” idea executable nights and weekends.

3. 3
   6:14 – 12:53

   Claude Projects as a business copilot: planning workflow and artifact-building

   Nabeel explains their operating system: a Claude Project containing many chats that generate living documents (business plan, checklists, timelines, drafts). Those artifacts are repeatedly re-ingested as context, compounding usefulness as the project evolves.

4. 4
   12:53 – 15:45

   Where AI helps (and where it fails): spatial layout vs. reasoning frameworks

   They highlight practical limits: LLMs struggle with floor plans due to weak spatial reasoning. In contrast, AI shines on frameworks—personas, pricing logic, competitive analysis—where textual reasoning and patterns are strong.

5. 5
   15:45 – 21:02

   Customer personas and event design to serve more than hardcore gamers

   Using AI, they build a persona matrix mapping game devotion/variety-seeking against introversion/extroversion. The output informs which events, formats, and programming can make the club welcoming to a broader audience than typical card-shop nights.

6. 6
   21:02 – 25:18

   Using AI to test viability: capacity, utilization, and ‘not a money pit’ math

   Before committing further, they use AI-assisted analysis to pressure-test whether the business can break even. The key is translating constraints—space, seats, hours, member behavior—into workable financial projections and utilization targets.

7. 7
   25:18 – 28:10

   Berkeley real estate and permitting: AI for confidence and second opinions

   They still hire lawyers/experts, but AI reduces the fear factor of navigating leases, permits, and local ordinances. They use models to interpret dense webpages, draft documents, and sanity-check advice—similar to getting a ‘second opinion’ workflow.

8. 8
   28:10 – 32:04

   From SaaS shopping to building their own: the AI concierge concept

   They first try off-the-shelf reservation and coworking management software but can’t find a fit for their ‘looking for gamers’ (LFG) matchmaking flow. The breakthrough is realizing they can build a simple database plus a chatbot interface to coordinate bookings and groups.

9. 9
   32:04 – 36:20

   Airtable as the business database for non-technical operations

   Andrew explains why Airtable is the backbone: it supports GUI-based operations, lightweight interfaces, and easier human intervention than a custom Postgres + front end. They also note AI changes the calculus over time, but Airtable accelerates setup and daily use now.

10. 10
    36:20 – 40:38

    Cataloging and merchandising games with AI: curated shelves + “Dewey Decimal” system

    They use AI to do heavy “librarian” work: creating curated retail categories and building a full taxonomy code (TLCS) that groups games by type and difficulty. This enables a discoverable in-store experience that would be prohibitively time-consuming manually.

11. 11
    40:38 – 43:42

    Demo: text-message AI concierge (Twilio + n8n) orchestrating real bookings

    Andrew demos a user texting a request (e.g., deck-building this weekend) and the system creating a booking request and reaching out to other members. The backend uses n8n workflows, Twilio for SMS, and agent prompts that route actions via Airtable tools (MCP).

12. 12
    43:42 – 48:54

    Experiences that wouldn’t exist without AI, plus lightning round lessons

    They close by emphasizing AI enables entirely new services rather than replacing staff—like dynamic matchmaking and large-scale cataloging. In lightning questions they share favorite games and practical prompting advice: reset approaches when stuck and add more context/tokens.