Features Now

See what your agent is doing without asking it.

Every agent session shows up on Now as a run: the epic, the task in hand, the test results, what it's waiting on you for, and what it did last. When it stops calling in, Now says so.

A run on the Now page. Each call the agent makes is a heartbeat: start_task, two checkpoints with checks at 38 of 42 and then 42 of 42, and finish_task. The run is on track. Then the calls stop. After 30 minutes without an update it says Quiet, with the time of the last one. After 4 hours it moves to Earlier. A new session starts, and a new run goes live on top.

The run card

One card per run, read in a glance.

Tasks done, working on, checks, waiting on you, blockers. The same five numbers every time, then the task list, the questions and the latest checkpoints underneath.

The Now page while an agent works. The run card shows sign-in and org onboarding on track: tasks done, the task in hand, checks going from 38 of 42 to 42 of 42, nothing waiting on you and no blockers, with each checkpoint in Latest.

What Now tells you.

  • Live, then quiet.

    A run is on track while the agent calls in. After 30 minutes without an update it says Quiet, with the time of the last one. After 4 hours it moves to Earlier.

  • Test results at each checkpoint.

    The agent reports passed and total at each checkpoint with log_progress. Failing checks show red on the card.

  • The task in hand.

    Working on shows the task it started last. The task list keeps it in view, with each task's status beside it.

  • Questions and blockers, in place.

    Open questions sit on the run with their defaults, and you can answer them there. Blockers say what's stuck and why.

  • Markers on tasks.

    A task shows when a default is in use, or when you answered differently and the agent hasn't picked it up yet.

  • Latest, newest first.

    Each checkpoint and handoff note: what the agent tried, what passed, what didn't.

Where it sits in the loop

Now is the Run step. The plan is recorded; this is where you watch it being worked, from any browser.

  1. 01PlanThe agent proposes the breakdown in your repo.
  2. 02Recordrecord_plan puts it in Beck as a document, epics and tasks.
  3. 03RunEach session shows up on Now while it works.
  4. 04AskQuestions come with a default, and the work goes on.
  5. 05ReviewYou accept the work, or send it back with a reason.
How the loop goes with an agent

See how your agent’s run is going on Now.

See plans

Free includes 50 AI credits. Recorded plans, runs and decisions use none.