About kutlhub
Real-time collaborative editing, always · in · sync.
Work stays where it belongs.
kutlhub provides a batteries-included UI that runs in your browser. Engineers use their editors and their AI agents. AI agents plug in through MCP tools or by co-editing files locally. Signals like flags, blockers, questions, and decisions get picked up from any of those surfaces. Every team member operates in the work style that fits their role, and the context stays with the work.
AI forward.
Most "AI-native" tools sell you a marked-up chatbot you could use directly for less. kutlhub encourages you to connect the smartest agents you can buy. The frontier will keep changing, and teams that let their work get locked into a house brand will fall behind. Your team connects the agents they already use, and can upgrade to the best agents available with no friction or context migration. Those agents plug in through the same system as the rest of the team, work on the same documents, and see the same signals.
What work looks like now.
The kutl technology.
What happens to my team's work if kutlhub shuts down?
Any team member can install the open source kutl client and keep a full local copy of every document. That's your team's off-ramp.
What happens to my team's work if kutlhub shuts down?
Any team member can install the open source kutl client and keep a full local copy of every document. That's your team's off-ramp.
Any team member can install the open source kutl client and keep a full local copy of every document in the space. That means your team's documents aren't locked into kutlhub: if kutlhub disappears, the documents are already on your team's machines. You keep editing the same files in the same editors. Running your own sync server is also an option, using the same open source tools.
How do AI agents actually plug in?
Through MCP tools or the filesystem. Agents bring their own inference; you pay the lab directly.
How do AI agents actually plug in?
Through MCP tools or the filesystem. Agents bring their own inference; you pay the lab directly.
Most agents integrate through MCP, calling structured tools through kutlhub's MCP server for reading changes, editing documents, and sending signals. Agents that don't speak MCP can also plug in by reading and writing files in a kutl space directly, and those edits sync to the team automatically. Both paths share the same identity model and the same signal primitives, so agents participate on equal footing with humans. Agents bring their own inference from whichever lab the team subscribes to. kutlhub doesn't resell model access or charge a markup.
What happens when two people edit the same thing?
All edits reconcile automatically. No merge conflicts to resolve manually.
What happens when two people edit the same thing?
All edits reconcile automatically. No merge conflicts to resolve manually.
Real collaboration is full of edge cases: two people editing the same paragraph at once, someone renaming a file while another is editing, a laptop going offline and reconnecting with changes, and an agent making different edits than a human at the same time. kutl handles all of them automatically and deterministically. The sync engine is built on established distributed systems theory, using the fastest CRDT text engine available by a wide margin. That's a structural guarantee, not brittle logic. The result is consistent state across the team, no matter how edits arrive.
What's kutl, and can I self-host it?
kutl is the open source sync protocol. kutlhub is the hosted product built on top of it.
What's kutl, and can I self-host it?
kutl is the open source sync protocol. kutlhub is the hosted product built on top of it.
kutl is an open source sync protocol and toolkit. The core is MIT/Apache-2.0 licensed; the reference relay is AGPL-3.0, all written in Rust. kutlhub is the hosted product built on kutl; we run the managed sync server and provide the browser UI, feeds, tasks, and invitations you've seen on this page. If you'd rather run the protocol yourself, self-host docs are at kutl.io.