Book a build call
Back to Help
Guide

Watch the work in the Run Center

4 min read

The Run Center is the page where you see what your agents did. It lists the recent runs your agents recorded, such as a workflow, a task or a turn with a visitor, with a plain summary of each one, so you can check the work without reading logs.

This guide walks the whole page. For the bigger picture of how your agents use more than one brain and what waits for your click, read One face, many brains: the Harness for owners.

Open the Run Center

Once you are signed in, open the menu at the top left of any dashboard page. Under Agents you will find Run Center, with the line What your agents did under it. The same menu section also holds Approvals, which comes up again at the end of this guide.

You can also go straight there: Open the Run Center.

The page opens with the heading Run Center — Your Agent Activity and the line Review what your agents did recently, with a plain summary of each run.

The four counts at the top

Four boxes sit under the heading:

  • Total Runs: how many recent runs the page found for you.
  • Successful: runs that finished.
  • Running: runs that are queued or still going.
  • Failed: runs that stopped with an error.

Only your own runs are counted. Another owner's work never shows up on your page.

A small chip at the top right says where the list came from. Live trace means the runs come straight from your agents' own trace record. Showing your recent agent activity means the page is showing the activity it has saved for you instead. Either way, you are looking at your own work.

If part of your history cannot be read at that moment, a yellow note at the top says so in plain words, and the page shows everything it can still read.

What one row tells you

Under Recent runs each run gets its own row, newest first, ten to a page. From left to right a row shows:

  • The name of the run in white, with its type and the agent that did it on the line below, for example a workflow name followed by your agent's name.
  • A status chip: green Success, red Failed or yellow Running. There are only these three.
  • The brain it used, as a small blue chip with the model's name. When the run did not record one, the row says No model recorded.
  • Duration: how long the run took, shown in milliseconds, seconds or minutes. A dash means no duration was recorded, for example while a run is still going.
  • Started: how long ago the run began, such as "12 minutes ago".

When you have more than ten runs, page links appear under the list.

Open one run

Each row has a View details button on the right. Press it and the row opens to show the Human Summary: one or two plain sentences on what the agent did in that run. When a run left no summary, it says No additional summary available for this run. Press View details again to fold the row closed.

The page shows the work as it was when you opened it. To see runs that finished after that, reload the page.

When the page is empty

A new account has no runs yet. In that case the list shows No runs yet and the line Once your agents start working, recent execution activity will appear here. Under it, one button gives you a real next step:

  • If you already have an agent, the button reads Try a turn with followed by your first agent's name. It opens that agent's Test Drive, where you try the agent yourself.
  • If you have no agent yet, the button reads Create your first agent and opens agent creation.

Either way, the button never sends you back to the dashboard.

The two buttons at the bottom

Two buttons sit at the foot of the page, whether it is empty or full:

  • Schedule an automated run opens the schedule page, where you set up work to run on its own.
  • Connect more business data opens the page where you add documents and web addresses your agents can learn from.

Runs and what waits for your OK

A run is a record of work that already happened. It does not ask you anything, and nothing on the Run Center changes a setting, sends a message or spends money. Work that needs your OK before it goes ahead waits on a different page: Approvals, listed in the same menu section with the line Things waiting for your OK. Open Approvals.

The Harness guide explains which of your agents' actions wait for your click and why: One face, many brains: the Harness for owners.

Back to the getting-started guides.

Still stuck?

The longer documentation covers models, voice, knowledge, and workflows in depth.

Open the documentation