Open Agents in the sidebar. The page carries one header with three sections under it. If your role can approve what agents propose, it opens on the first: what needs a decision from you. Everyone else lands on Agents.

Moving between sections keeps you on the same page, so the browser Back button leaves the Agents home rather than walking you backwards through the sections you looked at. The New agent button sits in the header on every one of them. See Building Your Own Agent.

Two sentences can appear directly under the section control, on every section, because they describe the state of the whole thing rather than of one list. The first is your AI credits, in the form Used about 8,000 of 10,000 AI credits this month. The count appears only for people who may see the workspace's AI credits (the billing.view permission, or everyone once member usage visibility is on). The second appears only when work is not being handed out: Agent runs are paused right now. Nothing is lost; they will resume automatically. When there is nothing to report, neither sentence renders at all.

The Inbox lists each proposed action as a card. It says which agent wants to do what, to which record, then one line with what would be sent or changed, such as To Dana Lee: Pricing follow-up or Stage: Negotiation (was Proposal), then what approving would do, and when it has to be decided by, in your own time zone: Decide by Thu 2:00 PM. If nobody does, it's skipped. Nothing on a card has happened yet. Approve and Reject sit on the card itself; open it for the whole draft, every value, and why the agent is asking.

A run that is waiting on you lists the same actions under Needs you, each with its decide-by time and the same Approve and Reject, so you can decide without leaving the run. The run shows the action and its deadline only; for the draft word for word, a note with a rejection, or several decisions at once, use the Inbox.

A decided card stays where it is and says what happened to it, such as Queued once you approve, while the change is carried out in the background, or Rejected. Once an approved change has been made it reads Done. Set the Inbox’s state filter to Approved or Rejected to see the last 30 days of decisions, which are still there after a reload. The card, the filters and deciding in bulk are covered in The Approvals Queue.

The Agents section opens with Results and the window it covers, This month. Four counts, and two of them open the inbox: Actions taken opens the queue, and Waiting on you opens it narrowed to what is still waiting. The two dispositions between them are counts you read, with nothing to press. The waiting count is the live queue figure the sidebar badge shows rather than a monthly one, so the tile and the badge can never disagree.

Under the four counts is a line per agent that did something this month. On a workspace where nothing has run yet the strip says so in words rather than painting four zeros: No agent activity yet this month.

Below that is the roster itself. Each row carries the agent’s name, how much it may do on its own in short form (Asks first or Automatic), when it last ran, and a switch to pause or resume it. An agent that cannot run right now says why on its own row, in a sentence, with the one thing that fixes it beside it.

Activity lists runs across all your agents. Each row shows the agent name, its state and what it ran on, a line on how it ended when it did not finish, and when it ran. Filter by state to focus on what matters, or land here filtered to a single agent from its row, with a chip you can clear to see everything again. When nothing has run yet it reads No runs yet. Open a row to read the run in a side panel, or follow the arrow at its end to the run inside its own agent’s Runs tab.

A run reads the same wherever you open it: in the side panel from Activity or from an agent’s Runs tab, on the run’s own page, and in Try it. At the top are the facts: the agent and its purpose, the state, what started it, when it started and how long it took. While a run waits on you, the time reads Waiting since 9:14 AM instead, and a test run adds Test run. On the run’s own page the facts sit in a column beside the readout, a Stop run button appears while the run is working, and the stages it went through replay under Pipeline.

Below the facts the readout runs in this order. A section with nothing to say is left out; What changed is always there.

A run its conditions skipped before it began has no steps and shows no line about them. Any other run with no steps to show says why, in one of three ways, and the first that applies is the one you read. Step details weren't recorded for this run. appears on runs that finished before agents recorded their steps, however old they are. It is a plain fact, not a fault: the state, any action waiting on you and the agent’s note still show. Next, Step details are kept for 90 days. appears on a run that recorded its steps, is older than that, and has none left to show. This run didn't record its steps. is different: the run finished and should have recorded its steps, and did not. That is a fault on our side, and we are alerted to it when the run ends.

Stopping a run is durable. It reaches work already handed out, prevents later claims, and reconciles any external effect that was already accepted. A stopped or finished run cancels every action still waiting on it, all at once.

The facts at the top of a run include one line on what started it, and where it can, the line links to what started the run:

• A record event: Started by: Company Acme was created
• An email: Started by: An email from Dana Lee: Pricing follow-up
• A form: Started by: A Demo request form from Dana Lee
• Someone running it by hand: Started by: Run by Sam Ortiz
• A test: Started by: A test on Acme
• A schedule reads as its schedule, in words.

An agent is told what started each run: the record it ran on and what changed on it, the answers on the form, the schedule, or who ran it. It reads an email’s words only when the agent was created by the owner of that mailbox; otherwise it is told only that an email arrived in someone else’s inbox. What it is told is treated as information, never as instructions, and is not stored with the run.

When something on our side stops a run, nothing is asked of you. If Laureo can try again, the run reads Retrying and says when: Didn't finish. Laureo will try again at 9:20 AM. If it cannot, the run reads Failed, we're on it and Didn't finish because of a problem on our side. We've been alerted. We are alerted, and no bell or email reaches you for it.

A run that stopped before Laureo recorded why reads just Failed, Partly failed or Timed out, with no sentence under it.

When the cause is something you can fix, the outcome says what happened and names the one thing that fixes it, as a link wherever there is somewhere to go:

A run stopped on purpose is not a failure. It reads Stopped and says why:

• This agent was archived, so the run stopped. Nothing was changed.
• You stopped this run.

The lists of runs show the state and one line on how it ended, with nothing to press. What you can do lives on the run itself, and only where your click changes the outcome: while the platform is retrying on its own, or after it has given up, there is nothing to press. A run that reads just Failed or Timed out, with no sentence under it, may still offer Try again or Run again, and one that reads just Partly failed offers Edit the agent. Where the cause is your monthly AI credits, your subscription, a permission or the agent’s own setup, the run offers the fix rather than a retry. Starting or stopping a run is for roles that can run the agent, and deciding an action is for roles that can approve.

Two of the eleven are about billing rather than about the agent. When the workspace has used its monthly AI credits, every agent pauses itself and each paused run points to the same fix, Add Extra AI Credits. Extra AI Credits start the agents again before the cycle resets, and they resume on their own when it does. When the subscription itself needs attention, the run points at billing instead.

A test run is not a twelfth state. Its readout carries Test run beside whichever of the eleven it reached, and in the Runs list a failed test reads just Failed. A test run offers nothing to press, not even Try again: you run a test again from Try it.

An agent page has three tabs: Build, Runs and Settings. Which one you land on is decided for you. An agent that is live, or paused after having run, opens on its Runs, because what it did is what you came for. A draft, or an archived agent, opens on Build, because what it is is the only thing there is to see.

Naming a tab in the link always wins over that rule, and opening one run always lands you on Runs with the run open, so a link from a notification never hides the thing it pointed at. The tab is in the address, so it can be linked, reloaded and shared; flipping between tabs replaces rather than stacks, so Back leaves the agent instead of walking through the tabs you looked at, while closing a run you opened does take one Back press.

If your role cannot approve, the Agents home opens on Agents rather than the Inbox. The Inbox tells you why in place of its queue, and a run shows the same sentence where Approve and Reject would be. The one person told about a proposal is the agent’s creator: when an agent you created proposes something, you are told in your notifications and in My Work, so you know it is waiting. Nobody else is notified. The proposal waits in the Inbox, where anyone who can approve, an admin by default, can decide it; ask one of them to, or to give you the permission.