Automation is the module nobody watches, because when it works there is nothing to see. Automation Insights watches it for you: how many runs succeeded this week, which workflows keep failing, which runs are stuck, which exceptions nobody owns, and which tasks a workflow handed to a person are now late — and it tells you which of those to fix first.
Written by the model, from the figures above
“Most runs are succeeding, but one workflow has failed on its latest run and will fail again on the next event — open its last failed run before anything else.”
If no provider is configured, the amber panel is simply absent. The figures above it do not change.
Every figure uses the same definition as the Automation Overview and the screen it opens, so the insights card, the overview and the run logs never give you three different answers.
Successful runs out of every run started in the last seven days, with today’s count beside it. A rolling week rather than all time, because a workflow fixed on Monday should stop dragging the number down by the following Monday.
A workflow counts as failing when its latest run this week failed or got stuck, or when it failed a quarter of at least four runs. The unit is the workflow on purpose: forty failed runs from one broken step are one problem, and a count of runs would make it look like forty.
A run still marked running thirty minutes after it started is counted as stuck. The insights read it as stuck without rewriting it — looking at a summary never changes the record underneath.
Open exceptions, with how many are critical and how many have no owner; workflow tasks past their due time; and the average time from an exception being raised to it being resolved over the last thirty days. These are the places an automation stops being automatic.
The card names one thing to do, and the order is fixed rather than scored: a failing workflow first, because it repeats itself on every event; then critical exceptions; then stuck runs; then exceptions with no owner; then overdue workflow tasks. Each recommendation links to the filtered screen where the work is done. Beside it sit a score and a separate confidence — a workspace with three runs this week gets a low confidence however clean its score looks.
This is the single most important thing to understand about Automation Insights, and it is the thing most “AI analytics” pages are careful not to say. Every figure on the page — every count, every total, every flagged record — is computed by an ordinary database query against your own data, with thresholds you control. A language model is used for exactly one job: turning that computed result into a paragraph a person can read quickly.
Counts, totals, ages and comparisons, run against your workspace under the same workspace scoping as every other screen. This step is deterministic: run it twice on unchanged data and you get the same answer twice. If the engine says nine invoices are overdue, nine rows matched.
What makes a request stale, a cost a spike, an entry large or a balance overdue is a number in your settings, not a judgement in a model. The defaults are listed below. Change one and the flags change with it, immediately and predictably.
Only then is a short plain-language summary generated from those facts. It is written on a background job so the page never waits for it, and it fails soft: if no provider is configured or the request fails, the summary is simply not there and every number on the page is unaffected.
The summary is stored with a fingerprint of the exact figures it was generated from. When those figures move, the fingerprint no longer matches and the summary is regenerated rather than served stale. This closes the failure that makes cached AI commentary dangerous: a confident paragraph describing last week, sitting directly above this week's numbers, with nothing on the screen indicating the two disagree.
Results are cached per workspace for a window you set — fifteen minutes by default — so the page is fast without being out of date, and a refresh action clears it when you want the current picture immediately.
These are the actual settings behind the flags on this page. They are per workspace, they are yours to change, and the defaults are chosen to be defensible rather than dramatic — an engine tuned to flag everything on day one is an engine nobody reads by day three.
| Setting | Ships as | What it decides |
|---|---|---|
| cache_minutes | 15 | How long a computed result is reused before it is worked out again. The refresh action clears it immediately when you want the current picture. |
| modules_enabled | automation: true | Whether automation contributes to insights at all. Switched off, the page, the insights bar, the overview card and the mobile card go with it — the workflows themselves keep running. |
Every flag carries the way to act on it. A finding is not a number on a dashboard you then go hunting for — it is rendered with a label and a link to the screen where the work is done. An insight you cannot act on from where you are reading it is a report, and you already have reports.
Stuck and failing are fixed rules today. Thirty minutes makes a run stuck, and a quarter of four or more runs makes a workflow failing, in every workspace. They are not yet settings in the table above — making them yours is in the ledger below.
A purchase-order approval workflow broke when an approver left. Nobody noticed, because a workflow that does not fire looks exactly like a quiet week.
The insights bar at the foot of the workflows screen reads 1 failing, success rate down to 74%. The recommendation is the failing workflow, with a link to its failed runs.
Every PO raised since Friday failed at the same step: the approval task was being assigned to a deactivated user. Eleven runs, one cause — which is why the card counts workflows, not runs.
The step is pointed at the role instead of the person, and the failed runs are retried from the run log. The eleven purchase orders reach their approvers before the morning is out.
With the failing workflow cleared, the recommendation moves on to the two exceptions nobody owned. They are assigned, and the week starts with nothing silently broken.
What AWRA OpsHub does today
More we can add to your workspace
Where we point you to a specialist
The middle column is work on top of an engine that already computes, caches and links. If your own stuck threshold or a per-workflow trend is the row that would change your week, tell us.
Triggers, conditions, approvals and actions — the workflows these figures come from.
Where a failed or stuck run becomes an owned, tracked exception.
Every module on one card, and the workspace-level summary across them.
Late orders, stale requests, cost spikes and how concentrated your supply is.
Revenue momentum, overdue invoices, average days to pay and who is slow.
The per-module switches, and what reaches a model.
Help Center
Run inventory, procurement, assets, sales, and field work with approved AWRA guidance for setup, migration, integrations, security, pricing, and support.
