Every founder and seed investor I talk to is troubled by the same question: where is it right to build a company that will not be run over by Claude a year from now.

There is an angle I think gets missed.

The markets most at risk are precisely the ones SaaS already conquered. Someone already did the hard work, convinced the customer to go digital, and built the bridge to software. Now an LLM can replace what sits on that bridge.

The other side is discussed less, and I think that is where the opportunity is. A huge part of the economy never went through that wave. Companies running on decades-old ERP, and entire departments inside modern organizations that were never digital-native. Agriculture, manufacturing, logistics, traditional industry. Very large markets.

The question is who will get there. It seems the frontier labs will not do it themselves. We saw this already with AWS, Salesforce, Google Cloud: companies built around a low-touch distribution model, usually relying on the enterprise sales organizations of integrators and distributors. It is not the same kind of company.

This week's announcements reinforce it. Both Anthropic and OpenAI set up multi-billion dollar JVs with Blackstone, Goldman, TPG and Bain, and are acquiring existing services and consulting firms. They are buying the bridge instead of building it. That will give them broad coverage of the generic middle, but it will not replace deep solutions that require specific understanding of a vertical or of a complex process inside a department.

This is where startups with an FDE model come in. They build the bridge themselves, and under the hood they consume the LLM exactly the way a startup today consumes AWS.

This is not competition. It is division of labor. The models and their big arms are the primitives. The startups are the focused distribution.

The digital gap of these markets was, until now, a reason to stay away. Now it is the reason to go in.