1. AI Startups Are Still Companies

• "Peel off the AI mask, and it's just a company. 95% is the same -- hiring, mission, team building."

2. Two Keys to Success: Trust and Persistence

• Trust: "People don't trust AI yet. Show your work, cite sources -- like showing your work in elementary school."
• Persistence: "People negative about AI haven't tried hard enough. The magic is in getting past 80% to the final 20%."

3. Portfolio Examples

• Day AI: Focuses purely on product reliability, not vanity metrics.
• Harvey: Won trust at the most demanding law firms -- trust trickles down from the top.
• Open Evidence: Started by admitting "I don't know" when uncertain, gradually improving accuracy.

4. AI Startup Characteristics

• Small teams with 10-100x productivity through AI.
• Distribution infrastructure already exists (internet) -- growth can be instant.

5. Vibe Revenue vs. Real Revenue

• "Magic moments (80/20) make users try, but they leave quickly. Solving problems end-to-end creates real revenue."

6. New Key Metrics

• Consumer internet metrics: DAU/MAU, daily/weekly retention
• Data flywheel: "Most founders claim it, but only 1% actually execute it."
• Margins don't matter yet -- token costs dropped 99%.

7. The Real Moat Is the Founder

• "The biggest moat is the founder. 'I am the moat. I won't stop until I succeed.'"
• Amazon and DoorDash wouldn't exist without founder persistence.

8. Pricing: Input vs. Output

• Output-based pricing likely becomes standard.
• New units like "hacker hours" can expand addressable markets.

9. Future Outlook

• Foundation models are more like MongoDB than AWS -- large but not winner-take-all.
• Real value is in the application layer.

"Building an AI startup is building a company. Build trust, push relentlessly. The real moat is you."