Claude Memory: what it remembers, how to use it, and how to manage it
Claude now remembers things about you across conversations. Here is how it works, what to tell it to remember, and how to keep it useful.
Claude Memory is Claude retaining information about you across conversations. Your name, your role, your preferences, your projects — things you have mentioned that Claude stores and applies in future chats without you having to repeat them.
This is different from your conversation history (which resets with each new chat) and different from Project instructions (which you write explicitly). Memory is what Claude learns about you naturally through your interactions.
What Claude remembers
Claude stores facts and preferences you share:
- Your name and role
- How you prefer outputs formatted ("I like bullet points" or "always use British English")
- Projects you're working on ("I'm building a marketing site for a B2B SaaS company")
- Decisions you've made ("we decided to use Next.js for the frontend")
- Preferences about Claude's behaviour ("don't explain things I already know" or "always show your reasoning")
Claude does not memorise entire conversations. It extracts the important facts and preferences and stores those.
How to use it effectively
Be explicit about what matters. If you want Claude to remember something, say it clearly: "Remember that I prefer concise responses" or "My company sells project management software to construction firms." Claude is better at retaining things you state as facts than things buried in passing.
Correct it when it is wrong. If Claude applies a memory that is outdated or incorrect — maybe you changed roles or your company pivoted — tell it directly: "I'm no longer working on the mobile app, I've moved to the platform team." Claude updates the memory.
Use it to skip setup time. The biggest value of Memory is skipping the first few minutes of every conversation where you explain who you are and what you need. Once Claude knows your role, your company, and your preferences, every conversation starts further along.
Managing your memory
You can view and manage what Claude remembers:
- View memories: Go to your Claude settings and look for the Memory section. You can see everything Claude has stored about you.
- Delete specific memories: If something is wrong or outdated, delete it.
- Clear all memory: If you want to start fresh, you can wipe everything.
Memory vs. Projects: when to use which
| Memory | Projects | |
|---|---|---|
| Scope | About you personally | About a specific body of work |
| Persistence | Across all conversations | Within the Project |
| Content | Preferences, role, context | Instructions, documents, files |
| Who sets it | Claude learns it from you | You configure it explicitly |
Use Memory for: things that are true about you regardless of what you're working on. Your role, your preferences, your communication style.
Use Projects for: context specific to a body of work. Your CS team's tone guide, your product documentation, your marketing brief.
They complement each other. Memory means Claude knows who you are. Projects mean Claude knows what you're working on.
For team admins
Memory is personal — each team member has their own. You cannot configure it centrally. What you can do:
- Encourage your team to tell Claude their role and preferences early. The faster Claude builds a useful memory, the faster everyone gets value.
- Remind people to update memories when things change. If someone changes teams or the company pivots, their Claude memory needs updating.
- Use Projects for shared context, Memory for personal context. Don't rely on Memory for things the whole team needs to know — that belongs in Project instructions.