Cortex is a hosted memory layer for AI assistants. It connects to ChatGPT, Claude and any client that supports the Model Context Protocol, so the same memory is available whichever assistant you open.
Most memory tools store everything they are given. Cortex does the opposite. Every write goes through a quality gate, and in production about 80% of proposals are rejected as redundant before they reach storage. Rejected proposals do not consume quota. What gets stored is broken down into typed claims, which makes two things possible: contradictions between memories are detected and kept visible, with an open or resolved state, instead of being silently overwritten; and every answer cites the memories it came from, so you can check where a fact originated.
A nightly consolidation merges similar episodes into more general memories. Search and recall are unlimited on every plan. Data lives on EU infrastructure, and export or deletion of your memories is self-service.
Cortex is built for people who work with more than one assistant and are tired of repeating themselves: consultants, researchers, small teams, developers building agents. Setup takes about two minutes. There is a 30-day free trial with no credit card; paid plans start at €0.99 per month.
