One shared memory for Claude, ChatGPT and all your AI assistants

Fezola's Memory is one memory shared by all your AI assistants: tell one assistant something, and the others connected to Fezola can recall it. Each assistant keeps its own slice of what you said, so you re-explain yourself in every new chat, every session, every tool. Fezola stores the memory on your account instead, and every assistant connected to your connector reads and writes the same one.

The problem it solves

• A new conversation starts from zero.
• Claude does not know what you told ChatGPT, and the other way round.
• The workarounds are manual: a paragraph about yourself pasted into each chat, markdown files, or a memory server you install and maintain.

How it works

Memory is part of the same connector as your mail, calendars and the rest, at https://api.fezola.com/mcp. Your assistants get five tools: remember, recall, update memory, forget and list memories.

• They remember by themselves. An assistant saves a short, self-contained statement when you say something lasting: a preference, a person, a decision.
• They recall before answering. Another assistant, in another session, asks Fezola what it knows about the topic.
• Fezola can also enrich it. What your assistants read and write through Fezola (emails, chats, calendar, attachments, drafts) can feed the memory automatically. Fezola does not go and fetch your messages on its own, and it does not see your conversation with the assistant, only what the assistant saves or reads through Fezola.

Sorted by project

Memories are filed by project automatically, with a common base called General for who you are and what you prefer. An assistant working on one project sees that project plus the General base. You can rename, edit or limit a project to an assistant.

What it stores, and what it refuses

It stores short statements: preferences, people, projects, decisions, lasting facts. It refuses secrets. Memories are handed back to assistants as data, never as instructions.

You stay in control

• See and edit. In the Fezola app, Memory lists everything: search, edit, delete one item, erase all, export.
• Switch it off. Memory is on by default. Three switches in the Memory page control automatic enrichment, whether assistants can write, and whether each save asks for your confirmation (off by default).
• Per assistant. On the connector page, each assistant has a Memory tick, and you can limit it to one project.
• Encrypted at rest, kept until you delete it, erased with your account. It is not end-to-end encrypted, because the server holds the key.

How it differs from Claude's or ChatGPT's own memory

Claude and ChatGPT each have a memory of their own, and each knows only its slice. Fezola's memory sits outside both: it follows you across tools, accounts and sessions. Other cross-assistant memory tools exist, usually something to install or host yourself. Fezola's is ready to use, and sits next to the mail, calendars and ad accounts you already connected.

Try it

Connect Fezola to Claude (see how to add a custom connector to Claude) or to ChatGPT, tell one assistant something worth keeping, then ask the other. To limit what each assistant sees, read choose which tools Claude or ChatGPT can see. For the safety side, see is it safe to give an AI access to your email?.

New to MCP? Start with what is MCP, the Model Context Protocol.

See also Claude memory vs ChatGPT memory and how a memory MCP server works.