With memory, your AI Assistant gets better the more you work with it. As you chat, it picks up details about you and your work, like which clients you look after or how a project is billed, and uses them in later conversations so you don't have to explain them again.
Memory works alongside your instructions. Instructions are rules you write yourself and want followed in every conversation. Memory is what your assistant learns on its own as you work together.
How Memory Works
Your assistant saves a memory in two cases:
When it decides something is worth remembering. While you chat, it notices preferences, facts, or ways of working that will help in later conversations, and saves them.
When you ask it to. Say "remember this" or "remember that…", and it saves what you told it.
If you correct something it already remembers, it updates that memory instead of saving a new one. For example, if you told it a project is billed monthly and later mention it's switched to a fixed fee, it keeps only the newer information.
Asking It to Remember or Forget
You don't need any special commands. Just say what you want in your own words:
"Remember that I'm the main contact for all our retainer clients."
"Remember that our website redesign project is billed monthly."
"Forget what I told you about how that project is billed."
When you ask it to forget something, your assistant deletes that memory.
To see what your assistant remembers about you, ask it in the chat, for example: "List everything you remember about me."
From the list, you can ask it to change or forget any memory.
Where Memory Works
Your assistant uses and saves memories when you chat with it in Productive.
Your memories are private. They're shared only between you and your assistant, and they're used only in your conversations with it. Nobody else on your team can see them.
📌 AI Agents have memory too. Each agent keeps its own memories for each person it chats with, so nobody else who uses the same agent can see what it remembers about you.
Memory isn't used in triggers. When your assistant or an agent runs on its own, on a schedule, or when something happens in Productive, it doesn't read or save memories.
Memory vs. Instructions
Memory and instructions differ in who writes them and how your assistant uses them:
| Instructions | Memory |
Who writes it | You | Your assistant, as you chat or when you ask it to |
When it's used | In every conversation, in full | When it's relevant to what you're asking |
What's kept | Exactly what you wrote | Your assistant's own summary of what you said |
Use instructions for rules you always want followed, like tone or format. Let memory handle the details about your work that build up over time.
👉 To learn more about instructions, see AI Assistant: Custom Instructions.
Getting the Most Out of Memory
You could use memory to:
keep track of which clients prefer which report format
remind your assistant of naming conventions your team uses for tasks and deals
save context about a project once, instead of explaining it in every new chat
let it know who on your team handles what, so it knows who to assign or mention
note recurring dates, like when a client's monthly report is due