Your context, available to every AI you use. Connect once. Every conversation with Claude, ChatGPT, or Gemini draws from the same understanding.
Mandaire indexes your conversations across providers. A decision you worked through in ChatGPT last month is available when Claude needs it today. No export, no paste, no re-narrating the history.
Import your AI conversation archives once. Mandaire goes back as far as you have data. The reasoning you already did with any AI compounds into every future conversation.
No switching. No new interface. Mandaire connects to Claude, ChatGPT, and Gemini as a standard MCP connector. Keep using the AI you prefer. The context just arrives.
Your data is yours. Mandaire holds an index built from your sources: people, context, commitments, history. You can inspect what is there, add to it, and delete any part of it. Nothing in the index is permanent unless you choose it to be.
Mandaire does not read your conversations for us. The AI queries Mandaire; Mandaire returns context to the AI. Nothing in that exchange reaches a third party. The index lives on your infrastructure.
Installing Mandaire is not delegating your memory to an AI company. It is giving your own AI tools access to context that was already yours.
You need a Mandaire account and a compatible AI subscription. The install is four steps.
In Claude.ai (Pro or Max), click your profile icon and go to Settings → Connectors.
Click Add connector. Enter: https://mcp.mandaire.com. Claude will show the connector name and permissions for you to confirm.
Claude opens a Mandaire sign-in window. Use your Mandaire credentials. After you authorise, the connector persists across sessions.
Ask about a person you have been in contact with, a prior AI session, or a commitment from last month. Claude will pull context from Mandaire as part of the same response.
Mandaire is in private beta. We onboard personally: connect your sources together, run a first session, check back at day seven.
Building a product and want Mandaire as infrastructure? Developer docs, wire format, and API reference are at mandaire.dev →