Team Memory is what stops you re-explaining your business to Tooli every time you open a conversation. When Tooli learns something durable about how your organization works — the words you use, the conventions you plan by, a correction you gave it — it writes that down and reads it back at the start of future conversations. Memory is shared across your organization, so what one planner teaches Tooli, the whole team gets.
Use Cases
Teaching Tooli your vocabulary once — "we call channels divisions" — instead of in every thread
Making a correction stick, so the same wrong assumption doesn't come back next week
Recording planning conventions: which channel you mean when you don't say one, how your fiscal calendar is set up, which scenario is the plan of record
Giving new team members the benefit of context the rest of the team has already established
What Memory Is
Memory is a small knowledge base that belongs to your organization. Tooli reads it before answering questions about your business, and adds to it when it learns something worth keeping.
Two things are worth knowing up front:
Memory is shared; conversations are not. Your conversation history is private to you. Memory is visible to everyone in your organization and to every conversation they have. Treat it as a team document.
Memory is for how you work, not what your numbers are. Tooli deliberately does not store figures — it queries those live, every time, so an answer is never based on a stale number sitting in memory.
What Tooli Remembers
Tooli only writes something down when it clears three bars: it's still true in a year, it's stable rather than volatile, and knowing it would change an answer.
In practice that means:
Vocabulary — what your team calls things, and what to mirror back
Conventions — the channel you mean when you don't name one, how you scope a "total"
Configuration truths — calendar setup, product hierarchy, which scenario is the plan of record
Corrections — the single highest-value thing you can give it
Ambiguity traps — name collisions and defaults people assume without saying
What Tooli Will Not Remember
Live figures — metric values, plan numbers, counts. These change, and a number in memory would become the answer Tooli quotes instead of the one it looks up.
Large lists — style, colour, and location catalogues. It queries those.
Credentials of any kind.
Facts about named individuals.
Open defects or workarounds — a bug note outlives the bug and then misleads.
Anything you've asked it to forget.
Teaching Tooli
You don't have to manage memory for it to work — Tooli writes as you chat. But you can be explicit:
"Remember that…" — Tooli writes it immediately and confirms what it stored.
Just correct it. "No, we call that a division" is the strongest signal there is, and Tooli records both the fact and what to do differently next time.
"Forget X" — deletes it.
Seeing What Tooli Remembered
Memory activity appears inline in the conversation, so nothing happens out of sight:
Checked memory — Tooli read from memory before answering, naming the files it opened.
Saved N memories — Tooli wrote something. Expand it to see each change.
Each change is listed separately and labelled by what it did: Remembered for something new, Updated memory for a change to something already stored. Expand a row to read the full text.
Every row carries its own controls, and the group has Accept and Undo to answer all of them at once:
Control | What it does |
| Keeps the change and dismisses the notice |
| Opens the memory so you can reword it before it settles |
| Removes a newly created memory |
| Restores the previous version of a memory Tooli changed |
You never have to act on these. Left alone the changes stay — the controls are there for the times Tooli recorded something you would rather it had not.
Managing Memories
Go to Settings > AI & Agents > Memories to see everything Tooli has stored.
The Team tab holds your organization's memories. The Toolio tab holds a set curated by Toolio that applies to every customer — read-only for your organization.
The list shows each memory's path, a preview of its content, when it was created and last updated, and who updated it — Tooli when the assistant wrote it, or the person's name when someone edited it by hand. From here you can open a memory to read it in full, edit it, delete it, or click New Memory to write one yourself rather than waiting for Tooli to learn it.
Memories are kept as small markdown files. INDEX.md is the table of contents Tooli reads first, core.md holds the orientation facts, and topic files like calendar.md or vocabulary.md hold the depth.
Admins can switch team memory off for the whole organization with the Organization memories enabled toggle. Turning it off leaves everything stored intact — Tooli simply stops reading and writing — and it applies to new conversations rather than ones already open.
Controlling Memory Per Agent
Each agent in Settings > AI & Agents > Agent Hub has its own memory setting:
Setting | What the agent can do |
| No access to memory at all |
| Reads team memory, never writes to it |
| Reads, and records what it learns |
Use Read-only for an agent you want informed by the team's context but not adding to it.
FAQs
Can my teammates see what Tooli remembered from my conversation?
Yes. Memory is organization-wide by design — that's what makes it useful. Your conversations stay private to you, but anything Tooli writes to memory is visible to everyone on your team and shapes their answers too.
Does Tooli remember my sales numbers?
No, and that's deliberate. Tooli stores how your business works, not what it did last week. Figures are looked up live every time, so an answer is never based on a stale number.
Tooli remembered something wrong. How do I fix it?
Three ways. In the conversation, use Undo on the group, or Delete / Revert on the individual change. Go to Settings > AI & Agents > Memories, open the memory, and edit or delete it. Or just tell Tooli "forget X" in chat.
Why didn't Tooli remember something I told it?
Tooli only keeps what's durable and would change a future answer — a one-off instruction for the current analysis isn't stored on purpose. If it matters beyond this conversation, say "remember that…" and it will write it.
Does memory carry over between Tooli and other AI tools?
No. Team memory belongs to Tooli inside Toolio. It isn't shared with the Toolio MCP server or any AI client you've connected separately.
Who can edit memories directly?
Anyone can undo a write from the chip in their own conversation. Editing, creating, and deleting memories from the Memories screen is an admin function.
Is there a limit to how much Tooli can remember?
Yes — 200 memories per organization, shown as a counter above the list. Teams land well under it in practice, because Tooli rewrites an existing file rather than adding a near-duplicate. If you approach the limit, tidy up from the Memories screen.