Focus
We do not invest in AI broadly. We invest in three layers where a more capable model is a tailwind for the company that owns the layer, and a threat only to whoever merely wraps it.
01
Vertical Intelligence
An autonomous worker for one regulated job: medical coding, claims, underwriting. It does not sell software to the team. It is the team, priced against payroll.
- We look for
- One regulated function, run end to end, priced per decision
- We avoid
- Co-pilots and point features inside the incumbent's tool
- Representative
- Harvey, end-to-end legal AI
02
Physical AI
The layer every autonomous machine runs on: simulation, perception, data, fleet operations. We do not pick the winning robot. We back the ground all of them stand on.
- We look for
- Simulation, data, validation and fleet software many programs run on
- We avoid
- Single-robot bets and anything sold to a military buyer
- Representative
- Foxglove, data and observability for robotics
03
Agentic Commerce
As agents start to buy on our behalf, every purchase needs identity, authorization, and settlement. Own that layer and you hold a toll on the category, not a feature inside it.
- We look for
- Identity, authorization and settlement rails that agents and merchants both plug into
- We avoid
- Shopping assistants, checkout wrappers, single-platform plugins
- Representative
- Skyfire, identity and payments for AI agents
One test
Three layers, one question.
A claims department, a robot fleet and a payment network look nothing alike from the outside. Underneath they pass the same test: the capability is a commodity input, the value sits in what the input has to pass through, and a better model widens the margin of whoever owns it.