Building a Real-World Business with Andrew Mason & Nabil Hyatt

By: Andrew Mason, Nabil Hyatt Host: Claire Vo Source: How I AI (ChatPRD) Type: podcast

Summary

How I AI episode with Andrew Mason (founder of Groupon, CEO of Descript) and Nabil Hyatt (partner at Spark Capital). Used AI as a third co-founder to open Tabletop Library, a brick-and-mortar board game social club in Berkeley. Three workflows: (1) Building a business from scratch with Claude Projects as central brain — started from a simple conversational prompt, generated financial projections, space layout, real estate and permitting guidance (fed Berkeley ordinance pages as PDFs), landlord letter of intent, marketing PowerPoint deck; developed 3x3 customer persona grid mapping dedication (one game vs. many) against introversion/extroversion, with AI estimating persona percentages and brainstorming event calendar per group; documents uploaded back into the project to build context over time; (2) Custom “Dewey Decimal System” for board games called TLCS (Tabletop Library Classification System) — designed hierarchical classification in Claude (400s = Cooperative Games, 420s = Adventure Co-op), decimal indicates complexity/weight from .1 (lightest) to .5 (heaviest); automated categorization of hundreds of games in Airtable by giving AI the game name and TLCS rules; curated retail collections using Airtable AI with natural language descriptions (“Games about cats,” “Silent Strategists”); (3) Text-based AI concierge for player matchmaking — tech stack of Airtable (database with flexible free-text fields designed for LM readability), n8n (workflow automation backend/agent brain), Twilio (SMS interface); member sends text with preferences, agent identifies member by phone number, uses tool-calling to query Airtable, finds compatible players, organizes game group — all agent instructions written in plain English system prompt with no code.

Key Ideas Extracted

Claude Projects as business “central brain”: Every idea, question, and decision fed into one project with a feedback loop — ask question, get document, upload document back to build context for next conversation
AI fills knowledge gaps for non-experts: Tech founders with zero retail experience generated business plans, financial models, permitting guidance, and marketing materials through conversational prompts
3x3 customer persona framework: Map customers on dedication (one game vs. many) and social style (introvert vs. extrovert) axes — AI estimates persona distribution and generates event calendars targeting each segment
Custom classification system (TLCS): Designed a Dewey Decimal-style hierarchy for board games with category ranges (400s = Cooperative) and complexity decimals (.1 to .5) — solves the “wall of games” discovery problem
AI-automated bulk categorization in Airtable: Give AI the item name and classification rules, it assigns codes to hundreds of items — saves hundreds of hours of manual research
Database design for AI readability: Used flexible free-text fields in Airtable for things like “gaming preferences” and “availability” knowing a language model would need to parse them — design data structures with AI agents in mind from the start
Plain English agent instructions: System prompt written in natural language tells the agent what to do; agent figures out which tools/functions to call — no code required for the orchestration logic
Low-code AI agent stack (Airtable + n8n + Twilio): Combines database, workflow automation, and SMS for a complete agentic system — member sends text, agent queries database, finds matches, organizes game night

Notes

Published Aug 4, 2025 on How I AI (ChatPRD). ~10 min read. (Note: filename says 2025-10-06 but actual publication date is Aug 4, 2025)
Sponsors: Lovable, Persona
Andrew Mason background: Founder of Groupon, CEO of Descript
Nabil Hyatt background: Partner at Spark Capital
Business: Tabletop Library — board game social club in Berkeley, CA
Tools: Claude Projects (planning), Airtable (database + AI categorization), n8n (workflow automation), Twilio (SMS), plain English system prompts
Key concept: AI as “third co-founder” — enables passion projects that wouldn’t exist otherwise
Three companion workflow guides published Jan 8, 2026
Cross-references: AI-native entrepreneurship, physical business with AI backbone, agentic workflows, database design for AI

Raw Content

Re-scraped from ChatPRD 2026-02-16. Full article content captured in Summary and Key Ideas above.