Product documentation
Find the path that matches your setup
Start with Chat alone, run Community Edition, join a managed workspace, or prepare a licensed private environment. Each guide states exactly which path it applies to.
Start with your current job
Choose your setup
Compare direct Chat, Community Edition, managed, and private paths.
Install Chat
Find the current distribution channel and connect your first endpoint.
Install Community Edition
Choose Windows, macOS, Linux, or Docker and follow the complete installer guide.
Connect an agent
Create an API key and send a working OpenAI-compatible request.
Browse by job
Start Here
Choose a product path, install Chat, and confirm compatibility.
- Choose your Orchestris setupChoose between direct-provider Chat, Community Edition, a managed workspace, and a licensed Private Deployment.
- Install Orchestris ChatInstall a currently published Chat build and connect either a direct provider or an Orchestris server.
- Supported platforms and requirementsCheck the device, host, storage, and network requirements for the Orchestris path you chose.
Deploy & Configure
Run Community Edition or prepare a licensed private environment.
- Install Community EditionChoose a currently published native installer or Docker image for the free, single-node Community Edition server.
- Install Community Edition on WindowsInstall the Windows x64 package, initialize the first organization, and operate the per-user tray deployment safely.
- Install Community Edition on macOSInstall the Apple Silicon disk image or package, complete graphical first-run setup, and operate the server from its menu-bar app.
- Install Community Edition on LinuxInstall the x64 DEB, RPM, or portable archive, initialize the first administrator, operate the systemd service, and preserve its SQLite data and secrets.
- Install Community Edition with DockerStart the free single-node server in Docker, create the first administrator, and open hosted Chat.
- Prepare a licensed Private DeploymentChoose who operates the environment, collect the licensed handoff values, and validate the deployment before rollout.
- Configure and secure a deploymentSet the public server origin, validate startup settings, secure network access, and preserve recoverable state.
Administer
Configure models, onboard people, and control access and usage.
- Configure providers, models, and aliasesAdd an AI provider, discover or enter its models, verify inference, create stable aliases, and grant models to members.
- Manage users and invitationsChoose the correct role, invite people with their intended groups, monitor delivery and acceptance, or create and remove accounts deliberately.
- Configure groups, access, and quotasGrant models through groups, add daily budgets when needed, and understand which limit blocked a request.
- Single sign-on support statusUnderstand the currently supported sign-in paths and the present limitation on customer-configured identity providers.
Use Chat
Organize conversations and understand synchronization and privacy.
Operate & Support
Review usage, diagnose problems, and follow published releases.
- Review usage and operational healthUse the Admin UI to review organization or system activity, interpret partial cost data, export the active report, and investigate operational signals.
- Troubleshoot errors and collect diagnosticsCheck service health, capture safe correlation data, locate Chat and Server logs, and interpret current HTTP error behavior.
- Find release information and plan an upgradeFind the release information linked to a current download, check required actions, and upgrade with a recoverable plan.
API & Agents
Connect approved agents through the supported OpenAI-compatible API.
Common next steps
Configure providers and models
Add a provider, verify models, create aliases, and grant access.
Control access and quotas
Understand roles, groups, daily budgets, API keys, and rate limits.
Understand sync and privacy
Learn exactly which conversations sync and which remain on one device.
Troubleshoot an error
Use the health endpoint, error code, request ID, and logs to isolate a problem.
Still need help?
Include the relevant request ID and timestamp when contacting support.