The Workflow That Failed Quietly
A workflow that stops working looks exactly like a quiet week. What Automation Insights reads from your runs, how it decides a workflow is failing rather than unlucky, and why it always names the same thing first.
Every other module tells you when something is wrong, because something is visibly wrong: stock runs out, an invoice goes overdue, a ticket breaches its deadline. Automation is different. When a workflow breaks, the work it was doing simply stops happening, and nothing on any screen changes. The approvals that should have been routed are not routed. The reminders that should have gone out do not go out. It reads as a quiet week.
That is why automation was, until today, the one module in AWRA with no insights page. Every figure it needed was already recorded — every run, every failure, every exception, every task a workflow handed to a person — but nothing read them together and said what to do. Automation Insights does that now, on its own page, in the insights bar at the foot of every workflow screen, and as a card in the Android app.
What it reads
| Figure | Exactly what counts |
|---|---|
| Success rate | Successful runs out of every run started in the last seven days. |
| Failing workflows | A workflow whose latest run this week failed or got stuck, or that failed at least a quarter of four or more runs. |
| Stuck runs | Runs still marked running thirty minutes after they started. |
| Open exceptions | Open, snoozed or escalated — with how many are critical and how many have no owner. |
| Overdue workflow tasks | Tasks a workflow assigned to a person, still open past their due time. |
| Time to resolve | From an exception being raised to it being resolved, averaged over the last thirty days. |
These are the same definitions the Automation Overview uses, computed in one place for both. If the overview says two workflows are failing and the insights page says one, something is wrong with the software rather than with your reading of it — so the two are built so that cannot happen.
A failing workflow, not forty failed runs
The headline unit is the workflow, not the run. A purchase-order approval that assigns its task to someone who has left will fail on every purchase order raised after they left. Forty failed runs, one cause, one fix. A count of failed runs makes that look like forty problems and buries the one workflow that matters under the noise.
Two things make a workflow count as failing. Its most recent run this week failed or got stuck — because the next event will almost certainly do the same. Or it failed a quarter of at least four runs — because a workflow that succeeds three times and fails once, every week, is a workflow with a condition nobody has found yet. Below four runs a single failure is not enough evidence of a pattern, and the second rule stays quiet.
Reading a stuck run does not change it
A run still marked running half an hour after it started is counted as stuck. The insights page counts it without rewriting it: looking at a summary never changes the record underneath. The run logs are where you retry or cancel it.
Why it always names the same thing first
The page names one thing to do, and the order is fixed rather than scored. Some problems repeat themselves; others only wait. The ones that repeat go first.
-
A failing workflow
It fails again on the next event that triggers it, and the one after. Every hour it is left adds work to the pile.
-
Critical exceptions
Raised as critical by the workflow or by a person, and the reason someone set the severity.
-
Stuck runs
Work half done. Retry it or cancel it from the run logs so it stops holding a record in the middle of a change.
-
Exceptions with no owner
An exception nobody owns is an exception nobody is resolving. Assigning it is a minute’s work.
-
Overdue workflow tasks
The workflow did its part and is waiting on a person. The task list shows who.
Each step links to the filtered screen where the fix is made — the failed runs, the critical exceptions, the overdue tasks — rather than to a general page you then have to search. When nothing is wrong, it says so.
The number is not the model’s
Every figure on the page is an ordinary query against your own workflow records. A language model is used for one job only: writing a short summary of figures that have already been computed. If no AI provider is configured, or your organization has switched third-party AI off, the summary is simply absent and every number stays exactly where it was. The page also carries a score out of a hundred and, separately, a confidence — a workspace with three runs this week gets a low confidence however clean its score looks.
- Who sees it: anyone who can see workflows. It uses the same access as the Automation Overview, so nobody gains a view of anything they could not already open.
- Where it is: Automation › Overview › Automation Insights, the insights bar on every workflow screen, the Awra Insights page, and the Automation Overview screen in the Android app.
- Turning it off: an administrator switches it off for the whole organization under Settings › AI modules. The page, the bar and the app card go with it. The workflows themselves keep running.
What AWRA OpsHub does today
- A seven-day success rate across every workflow run, with today’s runs beside it.
- Failing workflows counted per workflow, from a failed or stuck latest run, or a failure share of a quarter over four or more runs.
- Stuck runs counted at thirty minutes without the record being rewritten.
- Open, critical and unowned exceptions, overdue workflow tasks, and the average time to resolve an exception over thirty days.
- One recommendation in a fixed order, linked to the filtered screen where the fix is made.
- The same figures as the Automation Overview, from one shared calculation.
- A page, an insights bar on every workflow screen and an Android card, behind the same access as workflows and one switch under AI modules.
More we can add to your workspace
- Your own thresholds for a stuck run and a failing workflow, set per workspace in place of the fixed thirty minutes and quarter-of-four.
- Connector and webhook health inside the insight card; today it sits on the Automation Overview for people who manage integrations.
- A per-workflow trend over weeks, so a workflow drifting from 99% to 90% shows before it is failing.
- Time saved by each workflow, estimated from its runs and the manual step it replaced.
Where we point you to a specialist
- We do not switch off, edit or re-run a failing workflow on our own. The page tells you which one and why; changing what an automation does to your records is a decision for a person in your organization.
Per-workspace thresholds and a per-workflow trend are the natural pair, and we would quote them together.
Anything above that you need, we can build for you
Everything listed above as something we can add describes what ships in the standard product today — it is a starting point, not a limit on what AWRA OpsHub can do for your organisation. Kenya's eTIMS integration and its maintained payroll engine are both in the product because clients needed them and commissioned them; neither appeared by itself, and the same door is open for whatever you just read about. One qualification so this is worth what it claims: a small number of things on this blog we deliberately leave to a specialist rather than build — a statutory ledger we will not sign our name to, a rule that would decide a tax question for you, a clinical or member-funds record that belongs in a regulated system — and where that is true the post says so in those words. Everything else is a scope, a timeline and a price.
The operational work, which is what most commissions actually are
An extra approval stage in a chain that does not match the standard one, a custom field set on employees or assets that only your sector needs, an expiry that has to block an order rather than send an email, a report your board asks for in a shape nothing produces, or a scanner or weighbridge feeding the goods-in door. These are the commissions we are asked for most often and the smallest ones we quote — and unlike a revenue-authority pipeline, none of them waits on a regulator.
The module-shaped additions, which are the ones readers ask for most often
A price list with real discount authority, a customer-facing quotation that expires, a bill of materials or recipe costing, a staff advance that is issued, acquitted and chased, a member or unit ledger, a matching rule that holds a payment. Each of these is a build rather than a setting, and each has been quoted before — a bigger piece of work than a custom field, with a written spec and a date instead of a roadmap slide.
The report, document or pack nothing currently produces
The board pack in the shape your board actually asks for, a donor or funder layout, an invoice or receipt template carrying what your regulator or your customer expects, a dataset the report builder cannot reach yet. Usually the fastest thing on this list to deliver, because the data is already in the system.
Systems, rails and hardware you already run
The accounting package, CRM, online store, core banking or custom database you intend to keep — connected through our API so a fact is entered once and appears everywhere it is needed. Plus the physical edge: a scanner, a scale, a weighbridge or a till peripheral feeding the door it belongs to.
How it works: you describe the requirement, we return a written scope, timeline and cost, and once agreed it is built into your environment and maintained as part of the product. Nothing here waits on a regulator or a published specification, which is why operational builds are the ones we quote fastest. Tell us the requirement that would otherwise rule us out — that is a better first conversation than a demo.
Tell us what your operation needsThe short version
Automation Insights reads every workflow run, exception and workflow task and answers one question: is the automation still doing the work you built it for? It counts failing workflows rather than failed runs, reads stuck runs without touching them, uses the same figures as the Automation Overview, and names one thing to fix first in a fixed order — a failing workflow before anything else, because it repeats itself. It is on by default, sits behind the same access as workflows, and an administrator can switch it off.
Four questions to ask about any automation tool
How would I know a workflow stopped working on Friday?
A good answer sounds like
A named screen that shows it on Monday morning.
What it actually means
A broken automation produces silence. If the answer is that someone will notice, nobody will.
Do you count failed runs or failing workflows?
A good answer sounds like
Workflows, with the runs underneath.
What it actually means
Forty failures from one broken step are one problem. A run count hides which one.
When is a run stuck rather than slow?
A good answer sounds like
A number of minutes.
What it actually means
Without a definition, stuck is whatever the person looking decides, and two people decide differently.
Does the health page agree with the run log?
A good answer sounds like
Yes, because they share one calculation.
What it actually means
Two definitions of failing mean two answers to the same question, and the team stops trusting both.
See what Automation Insights computes
The signals, the order it uses, and what it does not pretend to be.
Read the feature pageFrequently asked questions
Does AWRA OpsHub show which workflows are failing?
Yes. Automation Insights counts a workflow as failing when its latest run in the last seven days failed or got stuck, or when it failed at least a quarter of four or more runs. It is on its own page, in the insights bar on every workflow screen, and in the Android app.
What counts as a stuck workflow run?
A run still marked running thirty minutes after it started. The insights page counts it as stuck without changing the record; you retry or cancel it from the run logs.
What does Automation Insights recommend first?
It follows a fixed order: a failing workflow first, then critical exceptions, stuck runs, exceptions with no owner, and overdue workflow tasks. Each recommendation links to the filtered screen where the fix is made.
Who can see Automation Insights?
Anyone who can see workflows. It uses the same access as the Automation Overview, and an administrator can switch it off for the whole organization under Settings › AI modules.
Are the figures written by AI?
No. Every figure is computed from your own workflow records. A language model only writes a short summary of those figures, and if AI is unavailable or switched off the summary is absent while every number stays the same.
