Memory
An AI that remembers everything so you don’t have to
Every conversation makes your agent smarter. It remembers decisions, preferences, project context, and key facts — then recalls them exactly when they matter.
Why Memory
Conversations that build on each other
Conversations that build on each other
Most AI chatbots start from zero every time. Conduity agents carry context forward. Mention a decision on Monday and your agent still knows about it on Friday — or next month.
Automatic, not manual
You don’t need to “save” information or tag things. The agent extracts important facts, preferences, and decisions from natural conversation and stores them automatically.
Transparent and editable
Every memory is visible on the Memory page. You can see exactly what the agent remembers, edit entries that are wrong, and delete anything you don’t want retained.
Categories
What gets remembered
| Category | Examples |
|---|---|
| Fact | "The API uses REST, not GraphQL" · "Launch date is March 15" |
| Preference | "Prefers bullet points over paragraphs" · "Uses TypeScript" |
| Decision | "We chose Postgres over MongoDB" · "Going with the blue design" |
| Context | "Working on Q1 product launch" · "Team is 8 people" |
How It Works
Extract, store, recall
After each conversation, the agent analyzes the exchange and extracts structured memories. Not every message creates a memory — only information useful for future context.
Memories are stored per workspace with the extracted content, a category, a confidence score, the source conversation and channel, and a timestamp.
When you send a message, the agent searches relevant memories before responding. If it finds something useful, it incorporates it naturally into the response.
Managing
Managing memories
View all stored memories on the Memory page with filters by agent, category, search, and date.
Click any memory to edit its content. Useful when the agent extracted something slightly wrong or a decision has changed.
Remove any memory that’s no longer relevant. Deleted memories are gone permanently — the agent won’t reference them again. Review low-confidence entries and decide whether to keep or remove them.
Compare
Memory vs. Knowledge Base
| Memory | Knowledge Base | |
|---|---|---|
| Source | Extracted from conversations | Uploaded documents and URLs |
| Format | Short structured entries | Full documents, chunked for search |
| Creation | Automatic | Manual upload |
| Best for | Decisions, preferences, project state | Reference material, policies, guides |
| Retrieval | Exact match + relevance | Vector similarity search (RAG) |
Memory gives the agent context about you and your team.
Knowledge base gives it context about your domain and content.
Privacy
Privacy and control
Memories are stored at the workspace level. They’re accessible to all agents in the workspace but never shared across workspaces.
Your memories are never used to train AI models. They’re stored in your workspace’s encrypted database and retrieved at query time.
There’s no hidden memory. Everything the agent remembers is visible on the Memory page. If you don’t want the agent to remember something, delete it.
Plans
Limits by plan
| Feature | Free | Pro | Team | BYOK |
|---|---|---|---|---|
| Memory | Read-only | Full | Full | Full |
| Conversation history | 30 days | Unlimited | Unlimited | Unlimited |
An AI that actually learns from working with you
Memory is enabled on every plan. Full management on Pro and above.
Try It Free