Why cheaper AI can’t buy Standard Form hospital trust

This article summarizes an episode of The Generalists’s video series featuring Owen Kosman, co-founder and CTO of Standard Form.

Owen Kosman, co-founder and CTO of Standard Form/ Photo credit: The Generalists

In healthcare software, cheaper code does not make trust cheaper. Owen Kosman, co-founder and CTO of Standard Form, sees AI cutting the cost of building products.

But the hardest parts of the market stay exactly as hard: proving value, earning hospital confidence, persuading cautious buyers to change.

San Francisco helps founders who bring something to test

The gap between building and adoption shapes how Kosman thinks about San Francisco. The city is most useful when founders arrive with a product, a clear question, or proof that they have already done the work.

Without that, it can become an expensive place for vague meetings. The first weeks are most useful when founders treat them as a live test of how they build:

• Build before arrival so meetings are about real work, not hopes.
• Live near other builders to meet people faster and see their habits up close.
• Use the first weeks to test customer access, lifestyle fit, and market pull.
• Share work openly because trust improves the quality of advice and introductions.
• Watch how skilled users run AI agents, then copy what improves your own work.

As Kosman puts it, “Building while talking is a good way to say, ‘I’ve actually been building this for the past few months. Do you know someone who’s very good at the parts of the technology I want to build?’”

That approach lowers the risk of moving too early. San Francisco becomes a place to test fit, buyers, and work standards before spending more time, money, or immigration effort.

The strongest gain may come from watching peers. Seeing another founder run several agents can change how someone works faster than reading product guides.

As AI tools improve, differences in how people use them can become differences in company speed.

Indonesia’s population does not prove software demand

The same test becomes harder in large markets where user numbers can hide weak purchasing power. A big population does not automatically create a big software business.

Kosman questions the idea that Indonesia’s 300 million people make it an obvious target for venture-backed software. His concern is payment: if people and companies have limited spending power, a startup can gain users and still fail to find enough buyers.

“The Indonesian economy right now is not suited for software that’s being sold in the US, because [consumers and businesses] just don’t have the purchasing power… it should cater to what is actually in demand,” Kosman argues.

US healthcare has budgets and urgent problems

AI lowers the cost of getting to proof

Healthcare defenses come from trust and rollout