Claude Memory Arrives for Enterprise Plans: What It Remembers and How to Opt Out Per Chat

Deep Dive: What changed

On September 18, 2025, Anthropic's release notes announced that users on Claude Enterprise plans can now use memory. Two concrete capabilities are described:

1. Memory across chats. Claude can remember relevant context from your conversations. Previously, each chat was a blank slate — anything you wanted Claude to know, you had to restate or paste in every time. Now that context can carry over.
2. A memory summary. Claude generates a summary of what it has remembered. This is the visibility piece: instead of memory being an invisible black box, there's a summary you can refer to.

The release also introduces incognito chats, which exclude a conversation from Claude's memory. That's the opt-out mechanism at the individual-conversation level — you don't have to choose between memory being always on or always off.

The scope caveat

Anthropic's release note ties this to Enterprise plans. If your team is on a different plan, the source material doesn't say anything about availability for you. Check with whoever administers your Claude account before assuming you have it.

How to use this in your day-to-day work

The source doesn't provide setup steps, so we won't invent any — but the two capabilities suggest a practical way to work:

1. Let recurring context accumulate in normal chats. The stuff you currently re-type at the start of every conversation — your role, your team's standard formats, ongoing initiatives — is exactly the "relevant context" memory is designed to retain. Over time, you should need less repeated setup per chat.
2. Use the memory summary as your audit tool. Because Claude generates a summary of what it remembers, you can check whether what it has retained is accurate and appropriate, rather than discovering a stale or wrong assumption mid-conversation.
3. Default to incognito for anything you don't want remembered. Sensitive personnel discussions, confidential deal details, one-off tangents that would pollute your working context — an incognito chat keeps that conversation out of memory entirely. The useful habit is deciding before the conversation starts whether it belongs in your memory or not.

What we don't know from this release note

The source doesn't describe how admins manage memory across an organization, how to edit or delete individual memories, or rollout timing beyond the Enterprise-plan scope. For those specifics, Anthropic's release notes and linked support articles are the place to check.

Note: claudedetails.com is an independent publication and is not affiliated with or endorsed by Anthropic. The facts above come from Anthropic's own release notes.

---

claudedetails.com is an independent publication and is not affiliated with, endorsed by, or sponsored by Anthropic. "Claude" is a trademark of Anthropic, PBC, used here for identification purposes only. Product details can change — always confirm specifics on Anthropic's own site before making decisions based on this post.