Map every agent, tool, employee and integration around one company brain. A live, inspectable view of everything your company runs — queryable from every AI you use, through MCP.
Connect your software, your teams and your AI agents. The Enterprise Brain maps all of it around one company brain — a live, inspectable graph of everything your organisation runs, knows and decides.
Leadership finally sees the whole picture: which agents are working, what they cost, who owns what, and how every tool and human connects.

Click any node for live metrics, model, token spend and run-rate — per agent. Spot what's working, and what's burning budget.

"What did I decide about the alpha launch last week?"
Brain queries: keyword + semantic + 2-hop graph traversal.
Real notes, real dates, linked to source. No hallucination.
The Fleece AI app is a thinking environment built on Brain. Every view reads and writes the same vault — what you do in Canvas appears in Graph, what an AI writes via MCP appears in Editor.
See your knowledge as a living graph. Pan, zoom, filter, focus.
The Fleece AI app is a thinking environment built on Brain. Every view reads and writes the same vault — what you do in Canvas appears in Graph, what an AI writes via MCP appears in Editor.
See your knowledge as a living graph. Pan, zoom, filter, focus.
Spatial whiteboard for thinking — notes, sketches, embeds, all on one infinite plane.
Block-based editor with [[wikilinks]], slash commands, transclusion, live backlinks.
Keyword + semantic + graph traversal — ranked together, returned in milliseconds.
Date-stamped pages with auto-rollups. Quick capture, scheduled reviews.
Tana-style fields and types for structured knowledge. Optional — markdown stays plain.
Markdown on disk. Rust engine in the middle. MCP tools at the top. Every layer is independent — swap, skip, or extend any of them.
Edit one config. Restart. 10 tools appear in Claude's tool list.
Same MCP config. Your code-agent now remembers your decisions.
VS Code agent reads your vault. Plans real work from real context.
Same folder. Same files. Open it side-by-side. Your humans stay happy.
First-class. Visual graph view, canvas, embeddings UI.
Speak MCP or raw JSON-RPC. Build your own agent on top.
Your files stay open and portable. We never invent a proprietary format. If Fleece disappears, your knowledge is still on your disk, in a folder Obsidian can read.
Source of truth is your laptop. Cloud sync is optional — and AES-256-GCM end-to-end. We literally cannot read your notes. We don't want to.
Brain is for AIs first — every API decision favors agents. But the data model is a knowledge graph you can browse, edit, and trust.
Yes. Plain .md + YAML frontmatter + [[wikilinks]]. Open the same folder in Obsidian — no migration, no surprises.
On your laptop. The vault is a folder of markdown files. Cloud sync is optional, opt-in, and end-to-end encrypted (AES-256-GCM).
Anything that speaks MCP — Claude Desktop, Cursor, Cline, Zed, Fleece AI App, custom agents. The 10 tools appear automatically in their tool lists.
Only for semantic search. Keyword and graph search work with no external dependencies. Embedding model is swappable.
Closed-source, proprietary. We may open select crates over time. The data format is open — your files are yours, forever.