Essay · 5 min read

The idea maze for AI startups

Chris Dixon · cdixon.org Free Advanced

A follow up that applies the idea maze to a wave many of you are riding right now. It shows how founders repeatedly chased the same category and kept hitting walls until the underlying capability finally became good and cheap enough. If you are building on today's AI models, this is a useful pattern for judging whether the wall has actually moved this time.

• Waves like AI attract repeated attempts, and most fail because the enabling tech is not ready
• The winning entry usually arrives right after a capability crosses a cost or quality threshold
• Study prior attempts in your category to see which wall just moved and which did not