Documentation
Documentation
Milieu runs AI agents inside your organization. Find the row that matches what you’re trying to do.
Find the right guide
| Goal | What it covers | Read |
|---|---|---|
| Try it | Stand up a working setup end to end | Getting started |
| Understand the ideas | Agents, skills, layers, sessions, and boundaries | Concepts |
| See how it fits together | The parts that run, the data, and the trust boundaries | Architecture |
| Use agents | Skills, knowledge, memory, messaging, and projects | User guide |
| Run it | The stack, daemon, gateway, API, milieu apply, audit, and troubleshooting | Operations |
| Deploy it | systemd on a virtual machine (VM), and Kubernetes | Deployment |
| Review security | Trust model, controls, and hardening checklist | Security |
| Look something up | Commands, environment variables, layout, manifest, and audit format | Reference |
Reading paths by role
- New user: Getting started → Concepts → User guide
- Skill or agent author: User guide and Reference
- Operator or site reliability engineer (SRE): Operations → Deployment
- Architect: Concepts and Architecture
- Security reviewer: Security and Architecture
Learn by example
The repository’s examples/ directory has four complete, runnable workspaces: a support desk, a contract-review team,
a SOC 2 audit, and a self-serve analytics team. You provision each one with a
single milieu apply, and each runs offline.
Scope
This site describes Milieu as it’s built today. The repository tracks planned work that hasn’t shipped yet.