Summary

Stancil argues that while building AI agents is the dominant startup strategy of 2026, most standalone agents are doomed because Anthropic, OpenAI, and Google will inevitably subsume them with general-purpose agents co-developed alongside their own frontier models. Instead of competing in the 'token path,' he suggests startups should build software designed *for* AI agents to use—tools, sandboxes, and infrastructure that robots need—rather than stuffing AI into their own products.

Key Insight

Instead of building AI agents that will inevitably be subsumed by the frontier labs' general-purpose agents, startups should build the tools and infrastructure that AI agents themselves will need to operate effectively.

Spicy Quotes (click to share)

• 8

  Once a lab decides a profit pool is theirs, they burn the surrounding economics down until the people sitting on it leave on their own.

• 7

  Do not let yourself become tempted by the allure of the token path, and instead ask yourself: If I had to build something without AI, what would it be?

• 7

  Perhaps, however, AI-native software will actually be software that is for AI.

• 5

  The robots will need better tools, designed for them. They will want notebooks to write down their ideas, sandboxes to play in, easy ways to connect to different sources of data, reliable ways to correspond with people, and utilities for making charts and presenting information.

• 4

  There are now far more businesses that are built around the automobile than companies that make the automobile.

Tone

satirical, cautionary, contrarian