Investment focus
Seven areas where we hold real conviction.
We invest with a thesis, not a checklist. Each area below is led by a partner who has built or operated inside it, and each thesis is written down before we write a check.
-
Artificial Intelligence
Applied models, agents and the tooling that turns research capability into dependable production software.
Our thesis
Capability is no longer the constraint; reliability is. The durable companies in this wave will be the ones that make model behaviour legible, testable and boring enough to put in front of a regulated buyer.
What we look for
- Teams who have shipped models into production, not only trained them
- A wedge that survives the next capability jump
- Evaluation and cost discipline built in from week one
-
Developer Tools
Products engineers adopt without a procurement cycle — bottom-up, opinionated, and loved before they are bought.
Our thesis
The best developer companies are distribution businesses first. If engineers will not adopt it without permission, no amount of enterprise motion will fix it later.
What we look for
- Organic adoption before any sales function exists
- A strong opinion the product refuses to compromise
- Founders who are the target user
-
Enterprise Software
Systems of record and workflow for industries where the software has been ignored for a decade.
Our thesis
The largest remaining software markets are the least fashionable ones. We look for sectors where the incumbent is a spreadsheet and the buyer has never been sold to properly.
What we look for
- Deep, specific domain knowledge in the founding team
- A workflow that becomes the system of record
- Evidence a real buyer will pay before the product is finished
-
Climate
Energy, grid intelligence and industrial decarbonisation where unit economics beat sentiment.
Our thesis
We fund companies that would win on cost alone if every subsidy disappeared tomorrow. Sentiment moves capital; economics move industries.
What we look for
- A cost curve that improves with scale, not with policy
- Founders comfortable with industrial sales cycles
- Measured data rather than modelled estimates
-
Healthcare
Clinical infrastructure, diagnostics and care delivery models that reduce cost while improving outcomes.
Our thesis
Healthcare rewards patience and punishes the demo. We back teams who have practised inside the system and know which fifteen minutes of a clinician’s day are actually available.
What we look for
- Clinical credibility in the founding team
- A reimbursement path understood at incorporation
- Retention measured in quarters of daily use
-
Fintech
Payments, capital markets plumbing and the regulated rails that move money across borders.
Our thesis
Regulation is not an obstacle to route around; it is the moat. The companies that build the licensing perimeter first tend to own the decade that follows.
What we look for
- Regulatory strategy designed before the product
- Unit economics that survive a rate cycle
- Teams who have operated inside a regulated institution
-
Infrastructure
Data, compute and security foundations that every other category quietly depends upon.
Our thesis
Every platform shift creates a new set of primitives that nobody owns for about eighteen months. We try to be early inside that window, repeatedly.
What we look for
- A primitive that other companies will build on
- Technical depth that is genuinely hard to replicate
- Willingness to stay unglamorous for years
Outside the thesis
What we don’t do.
Being explicit about this saves founders time, which is the scarcest thing they have.
Growth and later stage
Our last new position is typically Series A. After that we support the companies we already know.
Consumer social
We have no edge here and have stopped pretending otherwise.
Pure services businesses
Excellent companies, wrong fund. We look for products that compound without headcount.
Deals we cannot lead
We occasionally follow for existing relationships, but a party round with no owner is not a partnership.
Building something early?
Send a short note. Every message is read by a partner, and we reply within two business days — including when the answer is no.