WEBVTT

00:00:00.100 --> 00:00:03.915
The journey starts at eight twelve in the morning.

00:00:03.865 --> 00:00:11.462
Koru Access, a fictional internal business service, is producing intermittent sign-in failures.

00:00:11.462 --> 00:00:24.959
The service owner opens one operational incident case and records what is known: the affected service, the symptoms, the operating window, the estimated reach and whether a workaround exists.

00:00:24.959 --> 00:00:27.672
Try changing the impact.

00:00:27.672 --> 00:00:32.623
That detail will follow the incident into the response-level decision.

00:00:32.623 --> 00:00:36.353
The case gives changing work a stable home.

00:00:36.353 --> 00:00:44.655
New evidence and tasks can be added without losing the original report or pretending that the first description was complete.

00:00:44.810 --> 00:00:49.155
The incident case starts a published triage workflow.

00:00:49.105 --> 00:01:00.431
In this prepared example, the workflow reads a named service-status connection, retrieves the approved recovery runbook and finds two similar resolved incidents.

00:01:00.431 --> 00:01:06.427
A configured AI model can prepare a source-aware summary from that approved material.

00:01:06.427 --> 00:01:12.748
The sources and gaps remain visible, and a human reviewer confirms the incident picture.

00:01:12.748 --> 00:01:17.252
Select Gather approved context to reveal the prepared pack.

00:01:17.252 --> 00:01:21.877
The public page does not contact another system or model.

00:01:21.877 --> 00:01:32.431
It shows how repeatable triage could give the response team a useful starting point without turning a generated summary into an operational decision.

00:01:32.570 --> 00:01:37.647
Now the incident lead applies a configurable response-level decision.

00:01:37.597 --> 00:01:46.021
The example looks at business reach, service criticality, the available workaround and confidence in the evidence.

00:01:46.021 --> 00:01:52.016
Change an answer and the result updates with the score, matched rule and reasons.

00:01:52.016 --> 00:02:02.149
For the prepared incident, several departments are affected, the service is operationally important and no reliable workaround has been confirmed.

00:02:02.149 --> 00:02:07.209
That selects a major response and requires incident lead approval.

00:02:07.209 --> 00:02:11.129
This is not a universal incident-severity method.

00:02:11.129 --> 00:02:16.759
Your organisation would set its own inputs, thresholds and authority.

00:02:16.759 --> 00:02:22.510
The important point is that the result is explainable and stays with the incident.

00:02:22.660 --> 00:02:27.167
The decision result enables the work that is required now.

00:02:27.117 --> 00:02:37.440
The adaptive case plan contains a containment workflow, a stakeholder update, a recovery milestone and a discretionary dependency review.

00:02:37.440 --> 00:02:40.628
Select Activate response plan.

00:02:40.628 --> 00:02:48.807
The required activities become active, while dependency review remains available for evidence that has not arrived yet.

00:02:48.807 --> 00:02:53.379
This is where an adaptive case differs from a rigid checklist.

00:02:53.379 --> 00:02:59.334
It can preserve the possible work without forcing every incident through every task.

00:02:59.334 --> 00:03:08.504
Conditions and events make the relevant work available as the situation develops, and the case continues to hold the shared context.

00:03:08.640 --> 00:03:13.473
The case starts a mapped BPMN-based response workflow.

00:03:13.423 --> 00:03:28.032
This handles the work that should be coordinated consistently: preparing containment options, assigning specialists, reviewing a stakeholder update, waiting for human approval and then waiting for a recovery signal.

00:03:28.032 --> 00:03:32.888
Start the response workflow to see the linked run become active.

00:03:32.888 --> 00:03:37.867
The case and workflow have different jobs, but they remain connected.

00:03:37.867 --> 00:03:41.109
The case holds the changing incident.

00:03:41.109 --> 00:03:46.955
The process manages the agreed route and can wait without losing its place.

00:03:46.955 --> 00:03:57.210
In a real configuration, the mapped workflow version, roles, connections and approval boundaries would be chosen and tested by the organisation.

00:03:57.360 --> 00:04:05.652
At eight thirty-four, a new signal suggests that an identity dependency is experiencing unusual latency.

00:04:05.602 --> 00:04:08.233
Add that signal to the case.

00:04:08.233 --> 00:04:16.047
The evidence timeline retains the original report, the retrieved material and the earlier response decision.

00:04:16.047 --> 00:04:22.449
It now also makes dependency review available to an identity-platform specialist.

00:04:22.449 --> 00:04:29.951
The new evidence does not erase the earlier understanding, and the team does not need to restart the incident.

00:04:29.951 --> 00:04:36.191
The case changes the available work while preserving why the previous route was selected.

00:04:36.191 --> 00:04:46.161
This walkthrough uses prepared fictional evidence, but it demonstrates the practical value of combining evidence, sentries and discretionary work.

00:04:46.320 --> 00:04:55.941
The incident lead receives a compact decision pack with the current finding, the response rule, the evidence and a prepared stakeholder update.

00:04:55.891 --> 00:05:03.271
Pūnaha can organise those inputs, but it does not approve a consequential action on the person's behalf.

00:05:03.271 --> 00:05:05.875
Choose controlled reroute.

00:05:05.875 --> 00:05:14.190
The walkthrough records the human decision and shows that the prepared communication is approved for a configured continuation.

00:05:14.190 --> 00:05:23.496
A real connection could pass approved fields to an incident service or email route, subject to the organisation's controls.

00:05:23.496 --> 00:05:27.158
This public demonstration sends nothing.

00:05:27.158 --> 00:05:35.256
It keeps the accountable decision visible and makes the boundary between preparation, approval and action clear.

00:05:35.400 --> 00:05:42.539
After the containment route is approved, the response workflow waits for a correlated recovery signal.

00:05:42.489 --> 00:05:47.223
This durable wait represents a thirty-minute monitoring window.

00:05:47.223 --> 00:05:51.021
Select Simulate stable monitoring result.

00:05:51.021 --> 00:06:01.018
The successful sign-in rate rises, dependency latency returns to its expected range and the recovery signal resumes the workflow.

00:06:01.018 --> 00:06:07.313
The service-recovered milestone is reached only after that configured evidence arrives.

00:06:07.313 --> 00:06:13.769
The public page is not connected to monitoring, so the values are prepared and fictional.

00:06:13.769 --> 00:06:23.645
The design point is that recovery is supported by visible evidence and a named event, rather than being declared simply because a task was completed.

00:06:23.780 --> 00:06:27.053
Recovery is not the end of the work.

00:06:27.003 --> 00:06:31.072
Move the fictional clock forward forty-eight hours.

00:06:31.072 --> 00:06:37.760
A timer activates the post-incident review task and assigns it to the service-improvement lead.

00:06:37.760 --> 00:06:48.286
The task begins with the report, retrieved sources, case events, response decision, human approval and monitoring result already attached.

00:06:48.286 --> 00:07:00.088
The reviewer can examine what changed, update the runbook, begin problem analysis and assign follow-up actions without reconstructing the incident from separate messages.

00:07:00.088 --> 00:07:12.160
The full journey has combined an adaptive case, repeatable workflows, an explicit decision and human authority while keeping one inspectable history from report to learning.
