Stop Calling Them AI Wrappers. They Are Real Businesses.

There is a dismissive term floating around venture capital right now: AI wrapper. It refers to startups that build products on top of foundation models from OpenAI, Anthropic, or others, without training their own models. The implication is that these companies are not real technology companies. That they have no moat. That they will be crushed when the model providers build the same features.

This take is wrong. And it is going to age badly.

The Wrapper Critique Misunderstands Value Creation

By the wrapper logic, Salesforce is a database wrapper. Stripe is a banking wrapper. Canva is an Adobe wrapper. Shopify is a payment processing wrapper. Every application layer company is a wrapper around some underlying infrastructure.

The value is not in the infrastructure. The value is in the user experience, the workflow integration, the domain expertise, and the distribution. These are the things that make a product indispensable, and they have nothing to do with whether you trained your own model.

Where the Real Moat Lives

Data flywheels. Every time a user interacts with a vertical AI product, the product gets better at serving that specific use case. A legal AI that has processed 100,000 contracts understands nuances that a general-purpose model never will. That data advantage compounds over time.

Workflow integration. Building a chatbot that answers questions is easy. Building a product that integrates into an existing enterprise workflow, connects to their systems, handles their edge cases, and fits into their compliance requirements is hard. That integration work is the moat.

Domain expertise. Understanding how a specific industry actually works, not how it looks from the outside, is incredibly difficult to replicate. The startup founded by former radiologists building AI for radiology will always understand the customer better than OpenAI's general-purpose product team.

Why Model Providers Will Not Kill You

OpenAI has 400 million users. They are building for the broadest possible market. They will never build a product optimized for dental practices, or insurance claims, or construction project management. The market is too small for them, even if it represents a billion-dollar opportunity.

This is the same dynamic that has played out in every technology platform shift. The platform provider captures the horizontal market. The vertical players capture the industries. Both can be enormous businesses.

Our Position

We actively invest in companies building on top of foundation models. We do not penalize founders for using the best available infrastructure. We reward them for understanding their customer deeply and building a product that creates real value. The wrapper debate is a distraction from what actually matters: does the product work, and will customers pay for it?