Alcoven

Does Claude remember across chats and projects?

Blog · Published 16 September 2026 · Updated 30 September 2026

Yes, for you. Claude keeps a memory of your work that carries into new chats, and on paid plans it can search your past chats when you ask. Each project keeps its own memory, so what Claude learns in one project stays in that project. Everything Claude remembers is listed in your own Settings > Memory, and in a shared project your colleagues get the files and instructions, while your chats stay private unless you share them.

Here is how each part works, which plans have it, and where it stops, as Anthropic’s help pages describe it today.

How does Claude remember other chats?

Two features do the remembering:

  • Memory. As you chat, Claude saves what it learns as a set of topics: your role, the people in your work, how you like things written, project details and ongoing work. Anthropic’s own example: “Mention that a deadline moved, and your next conversation already knows.” You can also tell Claude to “remember this”.
  • Chat search. Ask “what did we discuss about the pricing page?” and Claude searches your earlier chats. The search shows up as a tool call, with citations linking back to the chats it used.

Memory also carries into Cowork tasks that run in the cloud. Cowork sessions that run locally on your computer don’t use it.

Feature Plans On by default?
Memory All plans (Free since March 2026) Yes on Free, Pro and Max; off on Team and Enterprise until an owner turns it on
Chat search Pro, Max, Team, Enterprise Yes
Incognito chats All plans Off until you start one
Memory import Free, Pro, Max (Team too, per the import article) You start it

Does Claude remember other chats in a project?

Yes, inside that project. Each project has “its own separate memory space and dedicated project summary”, so what Claude learns in one chat in the project is there for the next. On paid plans, chat search works inside a project too, limited to that project’s chats.

Memory keeps what Claude judged worth keeping, which isn’t everything. Anything every chat must get right, such as the brief or a decision, belongs in the project’s files or instructions: Claude uses those in every chat in the project.

Does Claude remember across projects?

No, by design. What Claude learns in one project “doesn’t carry over to others”, and your chats outside projects have a memory of their own. Search follows the same lines: it covers your chats outside projects, or one project’s chats at a time. If you keep each client in its own project, that separation is what you want.

How do you turn Claude’s memory on or off?

Open Settings > Memory:

  • Generate memory from chats turns memory on or off.
  • Search and reference chats turns chat search on or off.
  • Turning memory off gives two choices. Pause keeps what Claude has but stops it using or adding to memory. Reset permanently deletes all memory, project memories included.
  • Everything Claude remembers is listed under Topics. Open one to edit or delete it, or tell Claude in a chat what to change or forget.

On Team and Enterprise, an owner turns memory on for the organisation first (Organization settings > Capabilities), and after that each member manages their own. A few of these organisations still have the older version, under Settings > Capabilities, where the controls differ. Memory isn’t available to organisations with HIPAA, public-sector or custom data retention agreements.

Some things stay out of memory. Sensitive topics such as health, religious beliefs or politics are left out unless you turn on Include sensitive topics in memory, and some details, such as government ID numbers and financial account numbers, are never saved.

Does Claude remember deleted or incognito chats?

  • Incognito chats aren’t saved to your chat history or to memory, and chat search skips them. Start one with the ghost icon at the top right of a new chat outside a project; incognito isn’t available inside projects. Anthropic keeps incognito chats for 30 days (longer where an Enterprise organisation sets its own retention), and on Team and Enterprise they’re included in the organisation’s data exports.
  • Deleted chats. Deleting a chat doesn’t remove the memory Claude already made from it. Delete that memory itself in Settings > Memory.

Can you import your ChatGPT memory into Claude?

Yes. In Settings > Memory, select Start import. Claude gives you a prompt to paste into ChatGPT (or another assistant); paste the answer back and select Add to memory. Anthropic’s import article lists it on Free, Pro, Max and Team plans, on the web and in Claude Desktop (its getting-started FAQ names only Free, Pro and Max), and calls it experimental. It brings memory only: Claude can’t import your chat history from another AI. The whole move, projects included, is in How to move from ChatGPT to Claude without losing your projects.

Does Claude share memory with your team?

Anthropic’s help pages describe memory as one person’s: it’s what “Claude remembers about you”, on Team and Enterprise each member manages their own, and owners “can’t view or edit a user’s individual memories”.

A shared project, on Team and Enterprise, gives members the project’s files and instructions and lets them chat in it. Each person’s chats stay private unless they share one, and a shared chat is a view-only snapshot. The pages don’t say whether members of a shared project share its project memory.

Say you and a colleague work in the same shared Claude project. On Monday you settle the launch date in your chat. On Tuesday your colleague’s Claude has the project’s files and instructions, and your Monday chat stays private to you, so the sure way for the date to reach them is for someone to add it to the files.

Anthropic does keep some things for more than one person:

  • Claude Tag (public beta, Team and Enterprise) puts Claude in Slack channels. Its memory “belongs to the channel”, and anyone in the channel can read and correct it.
  • CLAUDE.md files in Claude Code hold a repository’s instructions and are “shared with your team through version control”. Claude Code’s auto memory stays on your machine. More in CLAUDE.md vs AGENTS.md.

The new projects in Claude Code (beta, Pro and Max) keep a project memory that every cloud thread reads, and the docs are clear that “a project belongs to one user”.

Claude’s memory also stays in Claude. A colleague on ChatGPT can’t open your Claude memory or your Claude project; to take memory elsewhere, Anthropic’s route is to ask Claude to write its memories out, then paste them into the other AI. ChatGPT draws the same lines from its side: Does ChatGPT remember across chats?

Where Alcoven fits

Claude’s memory is built for one person: you. A team also needs the project itself in one place, readable by whoever picks it up next, in whichever AI they use.

Alcoven keeps a project’s brief, documents, tasks and history, shared by the people and agents you add. Add the project’s link to Claude as a connector and your Claude reads and writes the project. A colleague opens the same project from ChatGPT, on plans where an MCP tool can write, or clones it with git to work from Claude Code or Codex. Whatever each AI changes comes back under that person’s name, for everyone on the project. Open a document in the Alcoven app and you see what changed since you last read it, and who changed it. Alcoven sits beside Claude: your memory keeps working for you, and the project is the part the team shares.

Free for one project and up to five people writing to it, no card. Start a project free

Sources

All checked on 30 September 2026.