User guideAI Engineering Platform
Back to siteSite

Overview

AI Engineering Platform is a workspace where AI agents execute the software lifecycle and people keep the decisions that matter.

A task moves through a defined path:

  1. Shape — a project blueprint creates the workspace; a task gets a description and acceptance criteria.
  2. Plan — the planning agent proposes steps, files and risks against a frozen context.
  3. Approve — a human approves the plan before code is written.
  4. Build — a development agent works in a disposable sandbox and produces a diff.
  5. Prove — tests, design checks and acceptance evidence are recorded.
  6. Review — a pull request is opened; a human approves the merge.
  7. Learn — delivery and quality metrics show what happened.

The platform does not hide these steps behind a chat box. Every transition is an event with an actor, an artifact and a timestamp.

Workspace overview

The workspace: health, projects and the next setup step.

Who it is for

TeamWhat they use it for
Product and platform teamsRepeatable delivery with approvals and an audit trail
Agencies and studiosOne workspace per client, with project-level access
Regulated teamsHuman gates, isolation and exportable audit evidence

What you need

  • A workspace (organisation) and a project.
  • A Git provider connection: GitHub App, GitLab, or a token for local development.
  • A model provider or a registered agent plugin.
  • A sandbox runtime (Docker locally, Kubernetes for a cluster deployment).

Principles

  • Plan before code. A task cannot start without a definition of ready.
  • Sandbox everything. The agent cannot touch the host, the network or another run.
  • Prove acceptance. The diff must carry evidence for every acceptance criterion.
  • Humans merge. No agent merges its own pull request.
  • Measure outcomes. Speed is reported next to rework, proof validity and escaped defects.

Where to go next