Building An AI Startup

• one perspective on important factors for building an AI startup are the 5 P's
  • positioning: aim at a vertical, avoid horizontals covered by big tech companies
  • product: build problem solving services, not tools
  • petabytes: scrape or find data to get over chicken/egg problems
  • process: sell ideas, help customers understand what problems are solvable and not through AI
  • people: hire technical, be technical

The Power of Data Network Effects

• Network effect is the positive feedback loop that improves user experience the more users on the platform
• Data network effect is positive feedback for AI products that improve with more data, while attracting more data providers as they improve
• "cold start" problem is like chicken/egg problem, and a pure AI company must have a data acquisition strategy
• data traps are interesting, you build a deliverable with tangible value to users, where useful data collection is a byproduct

Zero to One

• it takes forward thinking to create new technology, to progress vertically not horizontally
• technology enables vertical progress, the creation of new ways to do more, with less

What important truth do very few people agree with you on? A delusional popular belief hides a contrarian truth.

• 1990s, most industries were in recession, except for the tech industry
• Dot-com mania was from sept 1998 to march 2000; a Silicon Valley gold rush, a delusional popular belief
• the burst happened in March 2000; it scared most tech companies into the following dogma:

1. Make incremental advances
2. Stay lean and flexible
3. Improve on the competition
4. Focus on product, not sales. The only sustainable growth is viral

While this is the safe thing, the opposite principles are also valid:

1. It's better to risk boldness
2. A bad plan is better than no plan
3. Competitive markets destroy profits
4. Sales matter just as much as product