Interview, Podcast
How To Get AI Startup Ideas
- The YC AI startup school is scheduled for June 16th and 17th in San Francisco, offering travel cost coverage for eligible computer science undergraduates, graduates, and new grads, with limited application space.
- A high probability of success is predicted for founders persisting in the AI space rather than returning to traditional employment, contingent on committing to ideas that are difficult to ship quickly and avoiding "bandwagon" concepts.
- Founders are advised to leverage unique intersections of skills, deep domain expertise (such as PhD-level research or years at industry edges), or personal connections to access underserved markets, creating a "founder-market fit" that is difficult to replicate.
- Significant market opportunities exist in automating jobs currently performed by low-wage workers in fields like medical billing, auto loans, and dentistry, where AI can replace roles requiring only two to four weeks of training.
- Specific application areas for immediate automation include clerical tasks for police officers, reducing work time from hours to minutes, and replacing human consultants in robotic process automation (RPA).
- Founders are encouraged to search for remote analyst or clerk jobs on platforms like Indeed or identify outsourced low-wage functions to discover roles ripe for AI intervention.
- A shift is expected from B2B SaaS toward concepts capturing human imagination, such as new social networks or universal real-time translation tools, which may offer higher morale reserves for teams working on cutting-edge AI.
- Investment trends may shift toward "seed extensions" to allow more time for product-market fit, as companies like Zepto may rumor an IPO later this year and the AI agents being built are anticipated to be significantly larger than traditional SaaS.
- The current AI landscape can solve previously unsolvable problems requiring deep root-cause understanding, with specific examples including AI QA agents for front-end applications, circuit board verification, and LLM-powered back offices for dentistry.
- New startup concepts include AI voice agents for auto debt collection, people search tools for recruiting, search engines using vector databases for LinkedIn, and platforms for government contracting, all emphasizing deep technical execution over sales capability.
- Founders are urged to avoid social media-only research and instead gain "blinders" through undercover work in target industries, such as working as a biller or secret agent, to uncover hidden secrets and high-value ideas.
- Technical advancements such as vector database integration, fine-tuning models, and computer vision are enabling tools that improve RAG applications, coordinate logistics for truckers, and automate drive-through ordering.