Ship a feature from a plain-English request

Turn a plain-English feature request into shipped, reviewed work — with AI agents acting as product owner, architect, developer, and tester, and you approving at the gates that matter.

What you'll use

An SDLC Orchestration pipeline: a sequence of agent stages (product → design → build → test) with approval gates between them, so nothing moves forward without sign-off when you want it.

Steps

1. Go to AI Studio → Operate → Orchestrations → Start and choose the SDLC pipeline.
2. In the launch form, describe the feature in the free-text intent box and fill any required inputs the template asks for (e.g. target repo).
3. Launch. The Product Owner agent turns your request into user stories, the Architect designs the solution, and the plan lands at an approval gate.
4. Review the plan in the execution view. Approve to continue (or Reject with notes to send it back). Some gates let you attach a file or a structured answer that feeds the next stage.
5. The build + test stages run (often via a SiloLink coding session), open merge requests, and register their outputs as artifacts you can open or download right from the run.

Result

A traceable run: every gate records what was decided and by whom, every stage lists its artifacts, and the whole thing respects the cost limit you set. You stay in control at the approval points and let the agents do the legwork in between.