Welcome to OpenEnsemble
OpenEnsemble is an AI assistant you host yourself. Connect a chat model, create your assistant, and enable the skills you want it to use. Each account keeps its own conversations, files, settings, and connected accounts.
New accounts begin with one assistant. You can switch to a team of specialist agents later without deleting your existing work.
Start here
- Getting started — connect one model and get your first reply.
- Which model does what — chat, memory, personalization, and voice.
- Single assistant and ensembles — choose how to organize your agents.
- Troubleshooting — resolve common setup problems.
Terms you will see
| Term | Meaning |
|---|---|
| Assistant or agent | A named participant with a model, personality, and tools |
| Provider | The service or local model server that generates an answer |
| Role | An agent’s workflow instructions and default tools |
| Skill | A capability, such as email, calendar, coding, or web search |
| Coordinator | The agent that routes work to specialists in an ensemble |
| Project space | Shared project notes and files alongside separate conversations |
| Display dashboard | A dedicated browser view for a tablet, room, or status screen |
Where your data goes
OE stores your account data on your server. The providers and integrations you select determine which information is sent elsewhere. A cloud chat model receives the conversation context supplied to it; cloud speech providers receive audio or reply text. See Which model does what and Security model.
Use the Guide search to search every page, including pages you have not opened. On a phone, Contents switches between the page list and the current page; search shows matching pages automatically. Select a result to read it. What’s new lists recent changes.