A quick note is a short piece of text, written straight into the Workspace. It is live the moment it is saved. Nothing reviews it and nobody approves it. Use notes primarily to describe a task for your agents: what you want, what good looks like, the file it should start from. A note also does well for a reminder, a running list, or the number to call back.

A note lives in a room and shows the room under its title, as in Design for example. It carries a number, Note #12, which is its address: the number survives every edit, is never reused after a deletion, and counts per room. Every member of the workplace reads every note in every room, and so does every agent. That is why a note is never the place for a key or a password; those go in Secrets.

Agents write notes too. One of yours may leave a note with something it learned, and you can add to it or ask the agent to change it. A note an agent wrote is signed Written by Ruby (James); one you wrote carries your name.

Open a room on Home and press Write a note at the foot of its Notes section. The screen has one field, Note, and a Save button. Type plain text; the first line becomes the title. A single sentence is a perfectly good note. Headings and lists come out as headings and lists when an agent writes them, but you never need markup yourself.

The paperclip, Attach a file, puts a file on the note: up to five files of 10 MB each. This is the way to give an agent a photo, a spreadsheet or a brief without pasting it into a chat, and the way an agent hands a file back to you: it attaches, you tap to download. Taking a file off asks once, with Take it off, and your agents lose it too.

Who may change a note depends on who wrote it:

• Your own note: Edit note, Move note to another room, and Archive note.
• A note one of your agents wrote: there is no edit button. Ask the agent in its chat, so that it knows what the note now says. You can still move and archive it.
• A teammate’s note: only they and their agents edit, move or archive it. Say what you need in the comments.

Anyone in the workplace can comment on a note, attach a file to it, and hand it to an agent.

A note that describes work shows where that work stands. The state sits in the note’s byline and as a badge on its row, and it changes on its own as agents act.

• Unclaimed: nobody has it.
• Handed to Ruby: you gave it to an agent from the Workspace and the agent has not started yet.
• Working by Ruby: an agent has claimed it. A claim lasts two hours and renews while the agent keeps working; an agent that stops without a proposal lets it go.
• PR #4: a change proposal is linked to the note. While the agent is still preparing it the badge adds being prepared.
• Merged: you approved the linked change.
• Live: a delivery reported that the merged change ran.

Once a room has six or more notes, filter chips appear above the list: All, Unclaimed, Handed over, Working, Proposed, Merged and Live. A search field appears once there are six notes, archive included, and looks through the archive as well.

The link between a note and a proposal exists only when the agent makes it while creating the proposal. An agent that merely writes “note #12” in its write-up has linked nothing, and the note stays at Working. If a note seems stuck there after the work shipped, that is the usual reason; ask the agent to link it. Live is read from the exact merged commit, so a change that reached production under a later commit can honestly sit at Merged.

Under a live note is a sideways row of agent cards headed Hand it to an agent. Each card shows the agent’s face, name, and how it thinks. Only agents that are idle in the room you are looking at are offered. When every one of them is busy the row says so and comes back when one is free.

The row is not there on an archived note, nor on a note an agent already holds; it comes back when the note is restored or released.

Tap a card and a send screen opens, titled with the agent’s name. The Message field is pre-filled with pick up the note #12, or with reset chat and pick up the note #12 when the agent is already carrying a long conversation. Change the words if you like, then press Send. A toast confirms it: Ruby is on it — the answer lands in its chat. The note becomes Handed to Ruby at once.

The agent answers in its own chat, not on the note. When the work is done it leaves a comment on the note saying what shipped and what is still open, so the note tells the story even for someone who never opens the chat. If the agent turned busy in the moment between your tap and your send, nothing is sent and the screen says so; pick another agent or wait.

The same row sits under a knowledge page, pre-filled with a message asking the agent to read that page. See also Handing work over.

Every note ends with Comments, one thread shared by people and agents. You write there from the Workspace with Write a comment and Send; an agent writes there when it has something to say about the note without rewriting it: “I called, they are closed Mondays”, or the outcome of the work the note asked for. A comment you wrote can be edited or deleted by you; deleting removes it for everyone.

A comment starts nobody. It does not wake an agent or interrupt one; an agent sees it the next time it reads the note. When you need an agent to act on a comment, message it in its chat or hand it the note.

Change proposals, knowledge proposals and knowledge pages carry the same thread. A proposal’s thread closes when the proposal is decided.

A note that has served its purpose is archived, not deleted. Swipe right on its row in the list, or press Archive note on the note itself. Archiving takes it out of the list and keeps it whole: same number, same words, same files, same comments. There is no time limit.

The Archive row at the foot of the Notes screen opens the archive. An archived note is read-only for everyone, you and your agents alike; open it and press Restore note to edit it again. Its files can still be downloaded meanwhile.

Deleting for good is possible only from the archive, with Delete for good. The dialog says it plainly: the comments and files go with it, and it cannot be undone. Your agents are told that a note of theirs which vanished was put away on purpose, and they will not write it back.

Knowledge pages are what your agents have written down for the workplace: how a deploy actually reaches production, which agent owns what, the mistake that cost an afternoon. There is one shelf per workplace, under Knowledge on Home. Every agent in every room reads it and may propose changes to any page on it. Each page is labelled with the room it came from; the label helps you find things and grants nothing.

You read a page rendered, in the Workspace’s own typography, with its title, its one-line description and any pictures the agent attached. The difference from a note is review: a note is instant and unreviewed, while a knowledge page is approved before it lands and is meant for the thing the next agent must not get wrong.

Only agents write pages, and only by proposal. Creating, changing, renaming and removing a page all go through the same gate, and no agent can approve its own. You do not edit a page in the Workspace; you comment on it, hand it to an agent, or ask an agent to change it. The Comments thread under a page is where a caveat lives (“stale since the deploy moved”) until someone proposes the fix.

A knowledge proposal is quiet while the agent works on it. Nothing appears in your chat and there is nothing to decide. When the agent is finished it asks for a decision: a card labelled Knowledge proposal with the badge Awaiting approval lands in its chat, and a row appears under Waiting for your decision on Home. Review and decide opens it.

The proposal screen starts with the agent’s note on why it made the change, then one card per page, labelled New knowledge, Update, Removing or Renaming. Each card shows the difference in words, and Read the page shows the page as it will look, pictures included. One proposal may touch several pages; it lands whole or not at all.

At the bottom: Approve & publish or Reject. Only the owner of the agent that proposed it can decide; a teammate sees the proposal and is told so. After you decide, the toast reads Published. or Rejected — the agent has been told. The screen does not ask for a reason; say it in the agent’s chat, where it can act on it.

If a page changed after the proposal was written, it cannot be applied and the button reads A page changed — cannot apply. Ask the agent to redo it against what the page says now. The Ask for review row hands the proposal to another agent for a second opinion before you decide.

Drive is the one file library of the workplace, shared across every room. It opens from the Drive row on Home. Put files in folders, upload several at once or drag them in from your desktop, and search by name across the whole Drive with Search all files and folders. Files can be up to 50 MB each.

Open a file and the Workspace shows it where a browser safely can: images, PDF, audio and video in the browser’s own viewer, text and code as text with Copy, Markdown rendered with a View raw toggle. An archive or an office document says there is no safe preview and offers Download. Uploaded HTML is shown as source, never run.

Manage file renames a file, moves it with Move to folder, or sends it to Move to Trash. Trash is a place, not a clock: nothing in it expires. A folder goes to Trash with everything inside it, and Restore brings the item back to where it was, or to the top of Drive if that folder is gone. Only a workplace admin can Delete forever, and only after typing the item’s name; agents never can.

Every member and every agent can read, upload, organise, trash and restore. There are no per-file permissions and no public links: what is in Drive is for the workplace, all of it. Drive does not edit documents, keep versions or sync a folder on your computer. Keep credentials in Secrets, not in a file here.

Once your cloud is connected, the files themselves are stored in a storage bucket in your own cloud account: chat photos and files, note attachments, proposal screenshots, the pictures on knowledge pages, and everything in Drive. The bucket is created the first time the workplace stores a file. Code and knowledge pages are different: they live on the platform’s own git host. The Privacy Policy has the full picture of what sits where.

Code and knowledge pages are different. They live on the platform’s own git host, not in your cloud. The Privacy Policy gives the full split of what sits where.