Start with narrow AI use case that works today to fund expansive future vision
When building toward ambitious AI-powered future, start with narrowest use case that (1) current AI can execute well, (2) people will pay for today, (3) validates core thesis. Use revenue and learning from narrow execution to fund R&D toward bigger vision. This avoids 'build the future but starve getting there' trap. Focus on what AI can do well RIGHT NOW, not what it might do someday.
When to use
When you have expansive vision for AI but current capabilities are limited, when you need revenue to fund R&D not just VC money, when you can identify stepping stones from today's AI to tomorrow's vision, or when building consumer AI products. Best for founders with big vision but pragmatic execution mindset.
Don't do this
Promising future capabilities that don't work yet - kills trust when you can't deliver. Also avoid staying narrow forever and never pursuing bigger vision (just becomes lifestyle business). Don't use if narrow use case doesn't actually validate broader thesis.
2 Founders Who Did This
Vision: AI-customized movies, TV shows, all media personalized to each person. Reality today: AI-generated personalized children's books. Started with books because (1) current AI can generate good book content, (2) parents will pay for personalized books now, (3) validates personalization thesis before tackling video.
Started with narrow AI use case in 2019 (theme grouping with BERT at 50% accuracy plus human review) to fund the eventual vision of fully automated research analysis