AI CodeOps for platform and DevOps teams

Platform and DevOps teams are usually the ones asked to "roll out AI" across engineering. The hard part is not finding an agent. It is making AI work consistent, governed and affordable across many teams, repositories and Git hosts, without building the guard rails yourself.

The problem

Without a shared platform, every team picks its own assistant, its own model and its own review habits. Costs show up on different bills. Nobody can say which AI changes reached production, or which ones skipped review. When something breaks, the platform team gets the incident but none of the history.

How AI CodeOps helps

One way to run AI work

AI CodeOps runs agent sessions in the browser, on the platform's servers. Every session belongs to a project and runs a workflow. Built-in workflows cover common jobs: "Plan, Build, Review & Test", "Security-Sensitive Change", "Parallel Review", "DevOps Change", "Dependency Refresh", "Release Readiness", "Incident Response" and more. Built-in agents include an architect, builder, reviewer, tester, security auditor and DevOps engineer, and you can add custom agents. See agents.

Workflows can also run on a schedule, which suits recurring jobs such as dependency updates.

Guard rails you set once

Policies decide what needs approval, such as AI changes to protected paths or deployments. Every change passes a release gate: tests, the AI findings review and approvals. Approvals can require two people, and a change whose AI review could not complete is held. See quality gates.

Spend under control

Spend caps apply per workspace and per workflow, and each session's cost is tracked. Teams can run on the platform's Economy, Standard or Premium tiers, or Auto, which picks the tier per step. A workspace can also bring its own provider key, for example to use a contract you already have. See cost controls.

Supply chain and security

SBOMs in CycloneDX and SPDX format are tracked per project. Terminal sessions run in sandboxed pods, isolated from each other. See DevSecOps and SBOMs.

Deploys, incidents and DORA

The console records deployments, including rollbacks, and shows DORA metrics on the dashboard. Rollbacks are recorded here while your own pipeline does the redeploy. Incidents and runbooks live alongside, so the history of a change, its deploy and any incident it caused are in one place. See deploy and rollback.

Access and audit

Roles include owner, admin, developer and viewer, plus security admin and billing admin for scoped duties. The audit log records who changed what. Enterprise adds SSO (SAML or OIDC) and custom roles. Each change also has a signed evidence pack.

Any of five Git hosts

GitHub, GitLab (including self-managed instances), Bitbucket, Gitea and Forgejo all connect the same way, so teams on different hosts get the same workflows and gates.

What to try first

1. Follow the quickstart with one repository you own.
2. Set a workspace spend cap and a per-workflow cap for the workflow you plan to roll out.
3. Add a policy that requires approval for changes to a path you care about, such as deployment config.
4. Run the "DevOps Change" workflow on a small infrastructure task and follow it through the gate.
5. Look at the session cost, the audit log and the evidence pack, then decide which teams go next.

To automate setup, see the API docs. If you need to run AI CodeOps on your own infrastructure, self-hosting is available only as an Enterprise option through sales.