We fund the autonomy layer.
We back the infrastructure and vertical-AI systems that compound as models improve, not the wrappers the next release absorbs.
- Stage
- Pre-seed to Series A
- Focus
- Three layers of autonomy
- Approach
- High conviction, company by company
Thesis
Every infrastructure cycle ends the same way. Once the commodity gets cheap, value leaves whoever wrapped it and settles with whoever owns the layer it runs through. AI is there now.
It owns what the model cannot recreate: proprietary data, a system of record, a regulatory license, the distribution. A better model only makes it more valuable.
It wraps the model. The next release absorbs the feature and the traction goes with it. The demo was never the moat.
We fund the companies that get stronger.
What we back
Three layers a better model makes more valuable, not obsolete.
Vertical Intelligence
AI that runs the whole function
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.
Physical AI
The ground every machine stands on
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.
Agentic Commerce
A toll on every agent transaction
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.
The firm
Operator-led, framework-driven.
Layer H is led by Harrison Rolfes, a three-time founder and senior AI research analyst. He built the frameworks institutional investors use to underwrite the largest private AI companies, and applies them to companies early enough to matter.
The combination it takes to judge autonomous AI inside regulated markets. Most investors bring one. Layer H brings all three.
Underwriting now across all three layers.
Contact
Building, or backing, the autonomy layer?
Tell us who you are and what you need. Each note goes straight to the person who handles it, and every one gets a real read.