Reinventing software for the agentic era.






real-worldimpact
We ship AI that millions of people use and businesses depend on.
across more than 15 companies
average time from conception to launch
and counting...
blog

Replacing Sonnet as judge: Our experiments with Jev
We swapped Jev for Sonnet 5 on three closed-choice tasks and found it matched or beat the LLM, at a fraction of the cost. But batching small LLMs closes most of that gap without a new vendor. So when does switching earns its place?

Self-improving software: Survival of the fittest harness
We built a loop that lets software mutate and improve itself, then applied it to a coding-agent harness. Five generations in, it's more accurate and cheaper to run.

Building with agents: A conversation with the creator of OpenClaw
Five things we've been thinking about since our conversation with Peter Steinberger, the creator of OpenClaw.
openroles
We’re always on the lookout for exceptional talent. If we don’t have any relevant roles open but you think you’d be a good fit, please get in touch.