What Archon is trained on
The founder canon, retrievable.
Every chat turn pulls from 1,762curated excerpts — not Google search, not GPT's 2023 cut-off. The exact passages a senior coach would cite.
Paul Graham essays
1,762 excerpts · paulgraham.comTwenty years of the most-cited startup writing on the internet — every essay Graham has published from 2001 through today.
The foundational philosophy of recruiting users by hand, picking ideas you actually live, and surviving the first 18 months.
First Round Review
0 excerpts · review.firstround.comLong-form interviews with operators who have actually scaled companies — DoorDash COO, Stripe execs, Vercel CFO, Figma CPO, Vanta CRO, and 900+ others.
Tactical playbooks from people who built the things you're trying to build. Real numbers, real frameworks, no theory.
Scaling DoorDash, Notion pricing playbook, Stevie Case on enterprise sales
What this means for your answer.
When you ask Archon about pricing, it doesn't guess — it retrieves the relevant passages from Wealth, Startup = Growth, and the First Round pricing playbook, then synthesises an answer grounded in those.
When you ask about your first 10 users, it pulls Do Things That Don't Scaleand uses that lens against your live Stripe data. You get the tactic AND the founder context. ChatGPT can't do that.
Try a free audit →