AI actions

Explain, check and measure a process from one menu

Open AI actions on a process or a selection. Explain it in plain words, run a health check or an analysis, or start the optimization report, which measures waits and handovers in your event log.

AI actions open on the process in front of you.Run optimization reads the active event log, not just the diagram.Each finding names the step, its measured time and what to check first.The report changes nothing in the model. What to act on is your call.
Fixes and variants

Suggested fixes arrive as batches you keep or undo

Analysis findings come with fixes you can run. Fixes, clearer step names and TO-BE variants reach the model the way an AI draft does: highlighted, counted and reversible.

  • Analysis runs several review agents over layout, compliance and flow
  • Every run stays in the Analysis tab with its history
Say what should happen, in your own words.Parallel checks get a fork and a join, so onboarding waits until both are back.The newest group stays highlighted. Earlier ones settle into the model.Keep the draft, or undo every change in one click.
Routines

Some checks keep running after you close the tab

Change review, conformance drift and improvement opportunities can run on as read-only Wisp routines. Events from connected feeds pass a relevance check, and what touches your sites, suppliers or orders comes back as a proposal.

Connected feeds send events all day.Wisp checks each event against your sites, suppliers and open orders.What matters arrives as a proposal. A person reviews it before anything changes.
Preview · Object research

Research a site and keep the result in its dossier

Select a site, supplier or region in Horizon and start Research. Findings come back with sources and publication dates. The update waits as a draft until someone accepts it.

See Horizon
Pick a site, supplier or region on the map.Research runs on that exact object, with the focus you choose.Each example finding links to its source.The result waits as a draft. Nothing enters the dossier until someone accepts it.
How it works

From a question to a result you can act on

  1. Step 1

    Pick the target

    A whole process, a selection on the canvas or an object on the map.

  2. Step 2

    Run an action

    Explain, health check, analysis or optimization, each on the exact version in front of you.

  3. Step 3

    Read the result

    Reports and analyses stay in the AI sidebar with their history, so you can compare runs.

  4. Step 4

    Act on it

    Run a suggested fix as a batch you keep or undo, or leave the model as it is.

FAQ

Questions about AI actions

What data does the optimization report use?

The active event log for the process. Sections the log cannot fully support are marked limited, and without a log the report says that it is working from the model alone.

Does the report show individual employees?

It reports at process level, and worker data stays guarded. A review packet for data protection officers and works councils can be exported.

Can an AI action change my model?

Explain, health check, analysis and the report only read. Changes come from fixes or variants you start, and they arrive as batches you keep or undo.

Is there a risk and controls check?

It is in review and not released yet. It screens a process against a catalog of fraud and control scenarios.

How finished is object research?

It is in preview. The flow from Research to a reviewed dossier update runs end to end, and we are still qualifying it against live sources.

Next step

Run the report on one of your processes

Bring a process and its event log. We run the actions with you and go through the findings.