Operations Center

Findings and actions

The four tabs of the Operations Center, what you can do with a finding, how a proposed change moves across the Actions board, and what History records.

Findings and actions

What the agents measure, find and propose is split over four tabs of the Operations Center.

Examples before the first results

Until an agent has found, proposed or run something in this environment, its place on a tab shows examples of what will appear there. Each one carries an EXAMPLE tag.

  • Examples are never counted beside your own results. Once anything of your own is on a tab, the counts on that tab count only your own.
  • While a tab shows only examples, it says so at the top: What appears here after the first run · Examples, not your data.
  • Buttons on an example do nothing: there is nothing of yours to change.
  • An agent that is not available yet, such as Commerce, is always shown by its examples.

Signals

  • One group per agent, one card per signal.
  • Drawn from what the agents' runs measured in this environment.
  • See What a card shows.

Findings

What the agents found, grouped by agent.

  • Filters: All, High, Medium, Low (each with its count), Source (one agent), Status (open, closed or all) and a search by id or title.
  • Each finding shows its severity, its id (such as PERF-3F9A2C), its title, the pages it is about, when it was found and where its fix stands: Proposed in Actions, Waiting for approval, Fix in progress, Fix applied or Fix rejected.

The finding you select opens beside the list:

  • the agent that found it, its severity, the pages and when it was found;
  • Root cause: what the agent found;
  • the fields of its report, such as what was measured, the threshold and what usually causes it;
  • Expected impact and Revenue impact;
  • Proposed fix and, while a change for it is on the Actions board, where it stands there with Open in Actions.

What an agent does not record — its confidence, the expected impact, the revenue impact — reads Not available yet. No number is made up.

On a finding:

  • Dismiss marks it as not relevant.
  • Fix starts a run of an agent that can fix this kind of finding:
    • its estimate is shown first, and Start the fix starts it;
    • its proposals appear under Actions;
    • nothing changes until a person approves each one.
  • Fix is offered only where an agent can fix the finding. Otherwise it is greyed and says why, with the way on: Get the agent, Add the agent, Open its set-up, or Request it from the Laioutr team (nothing is charged).

The three agents of every project only read, so their findings are fixed by hand or by another agent that can fix them.

Actions

Every change the agents propose, on one board. Nothing is written without an approval.

ColumnWhat is in it
ProposedOpen findings an agent of the project can fix, with nothing proposed yet
Waiting for approvalChanges an agent proposed, waiting for a person's decision
In progressChanges approved and being applied
AppliedChanges applied; the newest three are shown, with how many came before and a link to History
RejectedChanges rejected, or approved and not applied
  • Assignee filters the cards by who holds them: who decided a change, or for whom the run that proposed it ran. Unassigned shows the rest.
  • Open a card to decide it:
    • Approve applies the change as the person approving, with their own access. Approving also needs the access the change itself needs.
    • Reject drops it. Nothing is changed.
    • Where a proposed change holds a hidden email address, the card asks you to type it.
  • Every decision is recorded in the organization's audit log and in History.

The outcome filters Applied with undo, Pull request and Guided steps describe how a fix reaches your project. Agents do not propose changes of these kinds yet, so they count only examples for now.

History

What happened in this environment, by day:

  • every agent run, Automatic or Manual, with who it ran for and what it found;
  • every change applied, approved or rejected, with who decided it;
  • every finding marked as not relevant.

Filter by type (Changes, Decisions, Runs, Settings), person and period (7, 30 or 90 days). Open run shows a run's status, when it started and finished, the credits it used, what it found and its steps.

Beside the timeline:

  • Last 30 days: changes applied, reverted, rejected, and agent runs.
  • Proof of work: Export report saves the entries of the period as a CSV file.
  • Recurring patterns: not available yet.
  • Retention: how long runs and decisions are kept is not decided yet.

A change cannot be reverted from History yet.

Hidden from you

Findings, proposed changes and measured values that need access you do not hold are hidden. The tab says how many.

Copyright © 2026 Laioutr GmbH