Private AI Platform
A private chat platform for language models running on your own GPU, reachable from anywhere.
The models stay home and the chat goes anywhere. Access is gated twice, and the chat connects to a knowledge vault, private web search and background agents that keep working after the chat closes.
- Type
- AI platform
- Platforms
- Web
- Status
- In daily use
- Built with
- Open WebUI, LM Studio, MCP, Cloudflare Access, Windows Server
- 01
Your models, your GPU
Chat with models running on your own hardware, from any browser.
- 02
Two gates
Cloudflare Access checks a one-time PIN at the edge, then the app asks for its own login.
- 03
Memory and private search
MCP tools connect the chat to a knowledge vault and a self-hosted SearXNG search.
- 04
Presets split by capability
A preset can search the web or run server tools, never both, closing the most direct path for prompt injection.
- 05
Work that outlives the chat
Long tasks run as background jobs, with a web console for jobs and schedules.
- 06
Built around one GPU
The chat stays responsive by keeping background agents to small, detached tasks.