The Truth About Building AI Startups Today

TL;DR

YC partners discuss that 50% of their summer 2023 batch are AI companies, but not due to a YC thesis – it’s an emergent signal from smart founders choosing where to build. They emphasize that boring, vertical, back‑office automation tasks are ideal AI startup ideas (e.g., government contract bidding automation), while generic “AI copilot” or “chat interface” ideas are often tar pits.

Key insights

• The high percentage of AI companies in YC batches is not because YC favors AI; it reflects what the most ambitious founders believe is the highest‑beta opportunity.
• Boring, repetitive information‑processing tasks (e.g., reading, summarizing, re‑entering data between systems) are a perfect fit for LLMs and are underexploited by applicants.
• “Tar pit” ideas in AI include AI copilots: easy to get inbound leads and revenue, but customers don’t actually use the product because they lack a specific use case.