Human-led. AI-operated.

What we work on

Vertical AI for industries where information asymmetry costs people most.

Horizontal AI tools get any given workflow about 30% of the way there. Closing the rest requires knowing how the industry actually runs — the language, the incentives, the workflows incumbents stopped questioning a decade ago. That’s where vertical AI compounds.

We pick industries where three things are true at once:

1. The professional class charges high fees for work that's largely mechanical. Pattern-matching, recall, judgment-on-rails. Real expertise lives at the edges; most of the day is the center.
2. The customer absorbs the cost when something goes wrong. Bad outcome, worse alternative, no leverage to switch.
3. The right software can make expertise portable. The win condition isn't a better dashboard. It's a product that knows what an expert would do and does it.

In practice, B2B and B2B-adjacent is where these converge — that’s where we hunt. Consumer stays on the table when the unit economics make sense; usually, they don’t.

What we’re not

Not a fund.

No LPs. No outside capital. No obligation to deploy on anyone else's timeline. Decisions get made on the merits.

Not an accelerator.

No cohorts. No demo days. No twelve-week programs. We pick what to work on carefully and build slowly.

Not anti-horizontal-AI.

Foundation models are the substrate. We build on top of them, not against them. Vertical AI without strong horizontal AI underneath doesn't work.