Does Claude remember your conversations now? What B2B teams need to know

Short answer: yes. Anthropic recently rolled out a memory update that unifies Claude's recall across Claude.ai chat and Claude Cowork, and it's on by default. If you've used Claude for a client project, then opened a new conversation a week later and had to re-explain who the client is, what you've already tried, and how you like your emails formatted, that specific friction is what this update removes.

TEG is an official member of the Anthropic Claude Partner Network. We build Claude Skills, Projects, and MCP integrations for B2B ops teams, and we help clients think through exactly this kind of update: what changed, what to turn on, and what to leave alone.

What actually changed

Before this update, Claude's memory reset with every new conversation. Each session started from a blank slate unless you manually re-pasted context, and Cowork and Claude.ai chat kept separate memory besides. Now, Claude carries relevant context forward automatically, and it does so across both surfaces at once. Tell Claude about a project in Cowork on Monday, ask a related question in a Claude.ai chat on Wednesday, and it already knows what you're talking about.

This isn't Claude logging a transcript of everything you've ever said. It's closer to how a good assistant works: it holds onto the durable, useful stuff — a client's name and situation, a formatting preference, a recurring task — and leaves the rest behind. Anthropic has also added controls for reviewing and managing what's been retained, and admins on business plans have visibility into how memory behaves for their team.

Why this matters for workflows, not just chat

Most coverage of this update frames it as a convenience feature for individual chat users. For teams running Claude-based automations, the more relevant story is what it does to setup overhead. A lot of Claude Skills and Projects exist specifically to compensate for the lack of persistent memory — prompt scaffolding, reference documents, and repeated context that has to be fed in every time a workflow runs. Native memory doesn't replace that scaffolding, but it does reduce how much of it you need for anything that depends on continuity between sessions rather than a single, isolated task.

Concretely, here's where it shows up:

• Recurring client or deal work. An account manager who talks to Claude about the same three accounts every week no longer has to restate who each account is and where things stand. Claude picks the thread back up.
• Meeting prep and follow-up. If you've asked Claude to help you prep for a recurring meeting before, it can build on what it already knows about the people and the history instead of treating every prep session as the first one.
• Style and format preferences. Teams that have corrected Claude's tone, structure, or formatting in past sessions should see less of that correction repeat itself.

None of this replaces a properly built automation. A Claude Skill that pulls live CRM data is still doing something memory alone can't: acting on current, structured information rather than recalled impressions. But for the day-to-day back-and-forth work that sits alongside your automations, this closes a real gap.

What to actually check before you rely on it

Before you treat this as "handled," there are three things worth doing:

1. Decide whether it should be on for client-facing work. Memory that carries context between sessions is useful, but if your team handles multiple clients in the same Claude account, confirm what's being retained and whether anything could bleed across contexts you'd rather keep separate.
2. Review what's already been remembered. Anthropic's settings let you see and edit what Claude has retained. It's worth a look, especially for accounts that have been in heavy use, so you know what's actually in there rather than guessing.
3. Don't confuse this with your data privacy posture. Memory is about continuity between your own sessions. It's a separate question from whether Anthropic trains on your data, which for Claude for Teams and the API is still no by default. We cover that distinction in more detail in our breakdown of Claude's data privacy policies for B2B teams.

What to do right now

If your team is already on Claude for Teams or Cowork, this update applies to you automatically — there's nothing to install. Spend ten minutes checking your memory settings, decide what should and shouldn't persist for client-facing accounts, and let the rest work in the background. If you've been holding off on Claude-based workflows because re-explaining context every session felt like too much friction, that's a smaller obstacle than it was.

Frequently asked questions

Does Claude remember past conversations now?

Yes. As of the recent update, Claude's memory is unified across Claude.ai chat and Claude Cowork and is on by default. Claude can recall relevant facts, preferences, and context from earlier sessions instead of starting from zero every time you open a new conversation.

Is Claude's new memory feature on by default?

Yes, it's on by default for accounts that have received the update, and it applies across both the Claude.ai chat interface and Claude Cowork rather than as separate memory stores. You can review, edit, or turn it off in your account settings if you'd rather Claude not retain context between sessions.

What does Claude remember about my team's workflows?

Claude can retain durable, useful facts it picks up in conversation: a project you're running, a client's name and context, a formatting preference, a recurring task you've asked it to help with. It's designed to remember the kind of thing a competent assistant would carry over between meetings with you, not to log every word you've ever typed.

Does this change whether Claude is safe for business data?

It changes what Claude retains between sessions, not Anthropic's underlying training policy. Claude for Teams and the API still don't use your conversations to train Anthropic's models by default. Memory is a separate, user-visible feature for continuity, and admins on business plans get visibility and controls over it.