AI application lifecycle

Cloud APIs support each stage of the Veryfront development lifecycle, from project source to execution history.

Build

Use the Projects API for source and branches, and the Agents API for agent definitions. Add retrieval through Knowledge, external access through Integrations, and model requests through AI Gateway.

Evaluate

Use the Evaluations API to run defined cases and inspect reports. Your release process determines which results are acceptable.

Deploy

Use the Deployment API to connect a prepared release to an environment. Check runtime readiness before treating it as available.

Run

Use the Execution API to inspect status, events, output and related runs. Agent, task and workflow runs can start through API requests, schedules or other triggers. Agent runs can also attach to conversations.

Operate

Use Observability for logs, metrics and traces. Identity and Access controls memberships and credentials. Billing and Usage exposes consumption and spending controls.

Improve

Use run evidence to reproduce a failure and add an evaluation case. Verify the change before another release. Cloud does not automatically change an agent or create an evaluation from a production failure.

See Cloud architecture for control-plane and runtime responsibilities.