Hire, prove, promote, review and retire every AI agent

Every Digital Employee in One3 goes through the same lifecycle, whether it is built in Agents Studio or rented from the Digital Employee Store.

A role is described or a template is picked. Studio generates the role, skills (as Actions), knowledge scopes, model policy and budget. Eval cases are generated automatically and the tenant adds golden tasks. Governance runs a risk assessment and the owner approves. The agent deploys in shadow mode (A1) with its own agent identity: it proposes, humans act, and agreement is scored.

Each skill is promoted separately to A2 or A3 when agreement and the Eval Coverage Score meet the threshold. In production, every run produces traces, cost records and outcomes. An incident, regression or drift triggers automatic demotion or a kill switch, an incident report, a fix and a re-evaluation. Otherwise, the agent gets a monthly performance review. Version upgrades pass an eval gate, run as a canary and are promoted or rolled back. At retirement the identity is revoked and memory archived per retention.

Agents Studio, shadow mode and K1/K2 kill switches are planned for W1; scorecards and monthly reviews for W2.

Every step on this page is an Action: a typed verb with a risk tier (R0 read-only to R4 irreversible or financial), a permission check and a reversibility class. Agents and humans call the same Actions, so the same approval rules apply whether a person clicks a button or a Digital Employee acts.