Yi EvalLive
Test an agent before launch and again after every change, and log, score and alert on its calls in production.
- Three ways to get started
- Datasets, scorers and experiments
- The release gate and CI
- CLI, API and security
DOCS
One set of docs per product, from setup to deployment.
Test an agent before launch and again after every change, and log, score and alert on its calls in production.
Before an agent calls a tool, runs code, touches a file or goes out to the network, Guard checks it against your policy and allows it, denies it or sends it to a person for approval.
Refactors and rewrites code in an isolated sandbox, verifies the change and hands you a draft PR; merging is up to you.
Model APIs, keys, sandboxes, VMs and billing. You can choose the region and machine type.
Sign-in, plans, top-ups, credit expiry and connected products.
Read account help ↗Next step