Self-hosted AI workspace
Illuminate your documents with us.
Build a library of your files once. Every chat — with Claude, Gemini, DeepSeek, Kimi, Mistral and more — starts already knowing it.
Three themes come up repeatedly. I've written them up so you can keep them:
What Lumius does
One library. Every model. Nothing forgotten.
-
Workspace Library
Upload PDFs, slides, spreadsheets, ebooks or plain notes. Lumius turns them into clean markdown and feeds them into every chat in the workspace.
-
Switch models freely
Pick a model per chat and change it mid-conversation. One key reaches every model your plan includes — no juggling vendor accounts.
-
Artifacts that come back
Summaries, reports and drafts arrive as artifacts. Save one to the library and it becomes context for your next conversation.
-
Your server, your data
Lumius runs on hardware you control. Documents and chat history stay with you, and your API key is encrypted at rest.
How it works
Three steps to a workspace that remembers.
-
Add your documents
Drop in files or paste text. Conversion runs in the background on your GPU, and each item shows how much context it takes.
-
Chat with any model
Your library is added to the conversation automatically, newest first. Lumius warns you before a model's context runs out.
-
Keep what's worth keeping
Edit an artifact freely, then sync it to the library when you're ready. Nothing in your library changes behind your back.
READS
- DOCX
- PPTX
- XLSX
- EPUB
- HTML
- JPG
- PNG
- WEBP
- Plain text
Bring your documents into the light.
Create a workspace, add your key, and start chatting with context.