Why Kythene
Doing the work is the easy half. Agreeing on it is the hard half.
Plenty of tools keep your AI's output and search it back. Kythene does the part that actually changes what your AI produces next: your team reviews the work down to the block, approves it, and the feedback lands back in the AI that made it - so the next result builds on what you agreed, not on whatever happened to be in the chat.
Review, not just recall
Your AI publishes a result; a teammate reads it, flags the one paragraph that's wrong and approves the rest; the notes go back to your AI over MCP. Nobody relays feedback by hand. That loop - human and AI working on each other's output - is the thing a memory tool can't give you.
How the loop works →Your AI actually uses it
A shared brain is only worth building if the instances read from it. Kythene connects over MCP, so recall and review are tools your assistant calls in the flow of its own work - not a wiki someone has to remember to open. The approved version is what it recalls.
What it connects to →See it, don't take our word
There's a real, populated workspace you can open and read - the timeline, a collection with its review, the memory behind it - with no sign-up. Read someone else's shared brain before you start your own.
Open the live workspace →Compared with the alternatives
vs Obsidian →
Obsidian is a superb home for one person's notes. Kythene is a different job: shared memory for a whole team and their AI, built on what the team actually agreed.
vs a memory server →
A memory server, or a weekend building your own, gives you store-and-search. The half that hurts is governance: what gets to change what your AI does.
Give your team a shared brain.
Free to start for one person, and free for the guests you invite to review. Bring your instances over MCP.