Why Artemis
You keep coding with your agents while Artemis works in the background. Its agents take tasks such as finding bugs, fixing performance issues and running Discovery Runs, using the models your team chooses.
Artemis runs as SaaS or on your own infrastructure, including air-gapped deployments with self-hosted models. There are three ways in. The Artemis CLI connects your terminal and coding agents such as Claude Code and Codex. CI jobs call the same CLI, so a nightly job can ask for a bug fix or a performance improvement. The web UI is where your team manages the agents and their work. Artemis has no built-in scheduler; the schedule is your CI job.
Code runs on a Runner on hardware you choose, with your own build, test and benchmark Commands. Artemis compares each changed Version with the Baseline, your unchanged code, on the same machine. Smaller open models running on your own hardware avoid per-token API bills, though you still run the hardware.
Artemis works best when a codebase has strong tests, evals and benchmarks, and it can help write them first. With AI writing more of the code, those checks matter more. Nothing merges without a developer's decision.