We are a team of 5 former founders and CTOs pushing the state of model capabilities for the leading frontier labs. We have built and shipped real products to real customers, and we bring that same bar to the environments and benchmarks labs train and evaluate on.
Refresh builds the simulation environments and evals frontier labs use to measure and push the frontier of what models can do, safely. An eval is only as useful as it is real, so we obsess over realism: environments that mirror actual computer work down to the last detail, with success measured the way it would be in production, not approximated.
We are hiring. If that sounds like your kind of work, see our careers page.