From the beginning, Relace was about building the right tool for the right job. We isolated the tasks that coding agents struggled with and trained specialized SLMs to do them instead. Starting with apply and embed/rerank, our models have made code generation faster, better, and cheaper for everyone.

We envision a future where coding agents can be built into all software systems -- the static SaaS pages of today will become malleable interfaces for the user to decide. At Relace, we want to build the models and infrastructure to make this possible.

Our team is based in-person in SF, where we work together. We're a team of ex-academics, founders, and hackers. If you're stubbornly optimistic and don't back away from hard technical problems, join us.