Claude now has memory. The real question is how it is scoped.

Does Claude remember previous conversations?

Yes. Anthropic's current help documentation says Claude can generate memory from chats and carry that context into new conversations. It stores memory as individual topics, updates those topics as you chat, and lets you explicitly ask Claude to remember something.

Claude also has past-chat search. On supported paid plans, you can ask it to find an earlier discussion, continue a project, or retrieve a specific detail. Search results appear as tool calls and include citations back to the source conversations.

Availability and defaults vary. Anthropic says memory is on by default for Free, Pro, and Max users on supported apps, while Team and Enterprise owners control whether it is available to members. Past-chat search has its own plan and organization requirements.

How Claude's project memory works

Each Claude Project has a separate memory space and its own project summary. That design keeps one project's context focused and prevents it from automatically blending with unrelated projects or ordinary chats.

Search follows the same boundary: Claude can search all chats outside Projects, or search conversations inside one specific Project. It does not treat every conversation in your account as one undifferentiated memory pool.

That separation is useful for client work, research, or any situation where context should stay contained. It can also explain an apparent miss: a detail remembered in one Project is not expected to appear in another.

Why can Claude still forget things?

Generated memory is selective. Claude saves topics it judges useful rather than replaying every message verbatim. A past detail may require an explicit chat search, may sit inside another Project, or may never have been promoted into generated memory.

Incognito chats are intentionally excluded from memory and past-chat search. Memory can also be paused or reset, and organization settings may disable it. Claude does not store certain highly sensitive categories, while other sensitive topics require a separate opt-in.

A long context window helps Claude reason over a large amount of material in the current request. Cross-chat memory and chat search solve a different problem: selecting durable context after that request is over. Neither mechanism guarantees perfect recall.

How much control does Claude give you?

Claude exposes its generated memory as editable Topics in Settings. You can inspect, change, or delete an individual topic, tell Claude to update memory in chat, turn search or memory off, and use Incognito when you do not want a conversation retained.

When Claude cites a past chat, you can follow that citation back to the source conversation. That provenance is valuable because it helps distinguish retrieved evidence from a generated memory summary.

How Fostera approaches continuity differently

Fostera centers continuity on a named Soul. It organizes retained context into episodic, semantic, and procedural layers and retrieves a relevant subset for each response. The system is designed for continuity, not perfect or exhaustive recall.

The memory browser lets you inspect, edit, and delete individual memories. Fostera Projects group related chats, files, instructions, and project-scoped memories while preserving the Soul's identity across that body of work.

The largest architectural difference is model portability. Fostera's memory and Soul identity live above the generation engine: auto mode routes among supported providers, and Premium users can manually choose available Claude, GPT, or Gemini engines.

Which should you use?

Choose Claude when you want Anthropic's native chat experience, cited search over prior conversations, Claude Projects, or direct access to Claude's newest capabilities. Its current memory product is real and user-controllable.

Choose Fostera when you want an ongoing Soul with visible memory entries, project-scoped continuity, relationship progression, and the option to change model engines without starting the relationship over.

You can also use Claude through Fostera. The comparison is less about whether Claude remembers and more about where you want identity, memory control, and project boundaries to live.