The category treats memory as a feature of the assistant — a context-window extender, a retrieval cache, a developer convenience.
Kortex treats it as belonging to the human, with the assistant as a tenant who is allowed in. From that one belief, several non-obvious commitments follow: portability is a hard requirement, the schema is open and exportable by default, and the product never makes you feel like you are renting your own knowledge back from a platform.
Connect a source once. Kortex listens for changes — new commits, new notes, new emails — and keeps the memory in sync, on its own.
Your connections are private to you by default. Encrypted in transit and at rest.
Whatever you ask through, Kortex injects the right memory at the right moment — only what the model needs to answer, only when it needs it.
Some memories are yours alone. Others belong to one project, with one teammate. Kortex makes the access boundary explicit — per project, per memory, per person.
The default is private. Sharing is a deliberate act.
We’re building it slowly and well. Drop your email and we’ll write when there’s something to show — no announcements, no marketing list, just one note when the beta opens.