Approach

AI is the foundation, not a layer

We believe the next generation of category-defining companies will be AI-native at their core. That belief shapes every decision we make.

Foundation, not a layerAI-native company: Product built around models, Data advantage, Operating model. AI added later: Feature on an existing product, Bolt-on tooling, Unchanged operations.AI-nativecompany01Product built around modelsData advantageOperating modelAI added later02Feature on an existing productBolt-on toolingUnchanged operationsFoundation, not a layerAI-native company: Product built around models, Data advantage, Operating model. AI added later: Feature on an existing product, Bolt-on tooling, Unchanged operations.AI-native companyProduct built around modelsData advantageOperating modelAI added laterFeature on an existing productBolt-on toolingUnchanged operations
Fig. 1 An AI-native company is built around models from the start. A company that adds AI later keeps its old structure.

Thesis

Our thesis

A company built around AI is different from one that adds AI later. The product, the data and the way people work all start from the same place.

Adding a model to an old product can help. But it leaves the structure unchanged. We look for companies where the structure itself is new.

That difference shows up in cost, in speed and in how much the company learns from every customer. It is hard to add afterwards, and easy to see once you know where to look.

Read the argument in why AI is the foundation, not a layer and what AI-native means.

How we evaluate

How we evaluate

AI-native by design

Would the product exist without AI? If a model can be removed and the company still stands, it is not what we look for.

Data advantage

Does use make the product better in ways others cannot copy? We test the claim. Data moats, real or hype.

Unit economics

Do inference cost, price and margin work at scale? See unit economics of AI products.

Team learning speed

How quickly does the team test, read results and change course?

How we operate

How we operate

We are operators first. We have built, scaled and run companies, so we work in the details with our companies: hiring, go-to-market, and data and infrastructure.

We prefer to be in the room when a hard decision is made. That might be a pricing change, a senior hire or a choice between building a model and buying one. We say what we think, and the team decides.

We also make fewer promises. We do not describe results we have not seen. We describe how we work and let companies judge us on it.

Read more about operators first.

Patient capital

Patient capital

We think in decades, not quarters. Patient capital lets a company make careful choices and take the time that good products need.

It also changes who we work with. We look for founders and owners who want a long relationship, and who would rather build something durable than chase a short-term story.

Founders can start at For founders, and owners and sellers at For companies. The four stages are described under Stages.

Contact

Building, running or selling an AI-native business?

Tell us a little about it. We aim to reply within two business days.