Your AI finally remembers. Everything. Automatically.
Memory Bank is the persistent memory layer at the center of Contained Evolution. Every conversation is evaluated, tagged, and stored. Every future conversation draws from it. You never re-explain yourself. Your assistant picks up exactly where you left off — every time.
Not storage. Judgment.
Every other AI tool either forgets everything when the session ends, or stores everything in a raw log nobody reads. Memory Bank does something different.
After every conversation, your assistant evaluates what happened. It asks: is there a fact here about who this person is? A decision they made? A preference they expressed? Something they intend to do?
If the answer is yes, that exchange gets distilled — tagged with hashtags, given a brief summary, and written to Memory Bank as a memory entry. Not the whole conversation. The thing that matters from it.
You don't manage this. You don't tag things manually. You don't decide what to save. Your assistant judges, and what it misses you can add yourself. The result is a Memory Bank that grows accurately over time — signal, not noise.
Memory that ages the right way
Not all memories stay equally relevant. Memory Bank manages this automatically through four tiers:
Hot — memories you've referenced recently or frequently. Surfaced automatically in every conversation. These are the things your assistant actively uses when it talks to you right now.
Warm — memories that come up occasionally. Available when the context calls for it, not pushed into every exchange.
Cold — older memories that haven't come up lately. Still searchable, still there, just not in the foreground.
Deep Space — archived memories from a long time ago. Each one gets a permanent Dewey code (CE.TOPIC.0001 format) so it can be retrieved precisely if you ever need it. Nothing in Deep Space is deleted unless you delete it.
Your assistant runs an automated process every night that moves memories through these tiers based on age and how often they've been referenced. A memory that stays relevant stays hot. One that's been quiet for months cools on its own. You never have to manage it manually.
See your memory organized back at you.
Memory Island shows you a live visualization of your Memory Bank — a force-directed graph where every memory is a node, every tag is a connector, and the size of each node reflects how often that memory influences your conversations.
Common patterns become large, prominent nodes. Rare connections become thin threads at the edges. Topics that come up constantly sit at the center. Things you've mentioned once live quietly at the periphery.
You can tap any node to see the memory it represents, edit it, or remove it. The map updates as your memory grows. It's not decorative — it's a real picture of what your assistant knows about you, and how it weighs that knowledge.
Tags that organize themselves.
Every memory entry gets tagged — by you, by your assistant, or both. Type #anything in any conversation and your assistant registers it as a user-applied tag. Your assistant also applies its own tags based on the content of what you discussed.
User tags always take priority over assistant tags on the same memory.
Over time, tags that appear frequently across many memories become anchoring categories — they get bigger in the Memory Map and the assistant leans on them as established context. Tags that appear rarely stay thin threads — they might be the beginning of something, or they might just be a one-off. The taxonomy corrects itself as your usage evolves. No maintenance required.
Your context, wherever you work.
With PASS, Memory Bank connects to external AI tools via MCP (Model Context Protocol). If you already use Claude.ai — or any other AI tool that supports MCP — your Contained Evolution Memory Bank becomes available inside it as a tool.
Your preferences, your decisions, your history — available to whatever AI you're working with, not locked inside CE.
This is the standalone path: during onboarding, you can bypass the full ecosystem entirely and go straight to Memory Bank. Skip the personal assistant creation, skip the Personal/Business sections, and use Memory Bank as a standalone memory layer for tools you already use. You can opt into the full ecosystem anytime from Settings.
Free vs PASS.
| FEATURE | FREE | PASS |
|---|---|---|
| Memory Bank storage | 15MB | 1GB |
| Sync across all devices | ✓ | ✓ |
| Automatic memory (PA-as-judge) | ✓ | ✓ |
| Memory Map visualization | ✓ | ✓ |
| Search, browse, edit memories | ✓ | ✓ |
| Hashtag system (user + PA tags) | ✓ | ✓ |
| Deep Space archiving + Dewey codes | ✓ | ✓ |
| MCP integration to external AI tools | — | ✓ |
| Cron automations while away | — | ✓ |