An applied AI lab.
We build software, hardware, and the businesses that run on both.
What we are.
Iterative is an applied AI lab pointed at the layer above the model. We build AI-native software, the hardware that runs on it, and the businesses that operate on top of both.
A model is not a product. The work lives downstream: in the interface that becomes a coworker, in the operation that runs differently because AI is in its core, in the device that learns. That's where we build.
Beliefs guiding our work.
- 01
Applied beats theoretical.
Models are research outputs. Products are the test. The hardest problems yield to deployed systems that make contact with reality and improve in cycles.
- 02
The model is not the product.
The product is the interface that becomes a coworker, the operation that runs 10× more efficiently, the device that learns. We build there, on top of whatever model is best that week.
- 03
The interface is the leverage.
Every great scientific instrument is a new way of seeing. The next decade's instruments are AI-first interfaces, and we build them.
- 04
Ship in the open.
Closed labs are slow. We publish what we learn, in public, while we build.
What we build.
Four thesis areas. Work spans all of them.
Agentic interfaces
Consumer and work software that behaves like a teammate, not a tool. Interfaces that take actions, not just suggest them.
AI rollups
Operating legacy businesses with AI in the core, not bolted on. The advantage is in the operation, not the model.
Hard AI problems
The unsolved infrastructure everyone else has to wait for: context, reinforcement learning, the plumbing the model layer left for someone else.
Specialized hardware
Physical products where AI is the capability, not a wrapper. Devices that learn.
A note from the founder.
The bet is that the most useful work for the next decade isn't training the model. It's everything downstream that turns the model into something a person actually uses: interfaces, operations, devices, businesses.
Iterative is the smallest team I could assemble to build across that whole stack: software, hardware, and the businesses that ride on both.
If you build with us, as a teammate, a partner, or a customer, that's the trade you're making.