THE MEMORY LAYER
A memory layer sits above the model.
A memory layer keeps what you know separate from the model that reads it, so you can change the model without losing the memory.
Last verified
What is an AI memory layer?
A store of what you know that is kept separately from the model that reads it. The model changes, the memory does not.
Why keep memory out of the model?
A model has no memory of its own. Anything it appears to remember was stored somewhere and put back in front of it. If that store belongs to the assistant, it goes when the assistant goes.
How does the assistant read it?
Through the Model Context Protocol, the open standard for connecting an assistant to an outside tool. You connect it once, and the assistant reads what you allow.
What AI Twin is
AI Twin is a personal AI memory platform. It keeps what matters to you as structured, searchable memory in the UK and EU, and works with Claude, ChatGPT, Gemini and other assistants while your data stays under your control.
The standard that makes it possible
The Model Context Protocol was released as an open standard around 25 November 2024 under an MIT licence.
Verified . Primary source.
MCP was donated to the Agentic AI Foundation under the Linux Foundation on 9 December 2025, co-founded with Block and OpenAI, with platinum members including AWS, Google, Microsoft, Cloudflare and Bloomberg. It is no longer any single company's protocol.
Verified . Primary source.
More on the Model Context Protocol