Can Claude remember our previous conversations?
How memory works in Claude
During an ongoing conversation, Claude retains the full history of that chat, so you can refer back to earlier messages without repeating yourself. This context window is large (up to 200,000 tokens for some models), meaning it can handle lengthy discussions.
When you start a new chat, Claude does not automatically remember previous chats. Each new conversation begins with a blank slate. This is by design to protect privacy and avoid mixing contexts.
Anthropic has introduced features like 'Projects' where you can group related chats and provide shared context, and 'Memory' (in beta for some users) that allows Claude to remember facts across sessions if you enable it.
- Within a chat: full memory of the conversation.
- Across chats: no memory by default.
- Projects: shared context for related chats.
- Memory feature: opt-in, remembers user-provided facts.
- You can always remind Claude of key details in a new chat.
Using Projects and memory
Projects let you upload documents and set custom instructions that apply to all chats within that project. This is useful for ongoing work like writing a book or managing a complex topic.
If you have access to the memory feature, you can tell Claude to remember specific things about you, like your name or preferences. You can view and manage these memories in settings.
Even without these features, you can manually paste important context at the start of a new chat to bring Claude up to speed.
Common mistakes
- Expecting Claude to remember your previous chat in a new session without using Projects or memory.
- Assuming memory is always on—it's optional and may not be available to all users.
- Forgetting that clearing your chat history or using incognito mode will erase context.
