Team rollout, change control, and incident handling
Roll out changes with an owner, a stop rule and a recovery plan.
In this lesson · A7 of 8
All lessons in this track →Progress for this track is saved in this browser only. No completion badge is issued. View your learning.
Start with the idea
Roll out changes with an owner, a stop rule and a recovery plan.
A new tool or data path changes what the workflow can do. Review that change before use. Keep a safe fallback and require evidence before restarting after a failure.
Rollback: Returning to a previous version; the restored behavior must still be checked.
See the example and check your understanding
A made-up teaching example
Version one only drafts. Version two adds sending and drops source IDs. When an unapproved send is attempted, pause the workflow and follow the established incident process. Use manual drafting while it is repaired.
Ask yourself
Does returning to the old version automatically approve restarting?
Show the explanation
No. Check the restored behavior and get the required decision from the responsible owner.
Start with these checks
- Name who can stop the workflow.
- Preserve evidence under the handling rules.
- Check recovery before resuming.
Use the different practice case below and check every requirement in its full instructions.
The video explains the main idea. Use the sources, instructions and checks below to complete the practice.
Captions appear here during playback.
Team rollout, change control, and incident handling
Roll out changes with an owner, a stop rule and a recovery plan.
Video playback is unavailable. Read the descriptive transcript below, go directly to practice, or open the video file.
Source-based teaching example · Fictional data · Reviewed 2026-10-03. Microsoft Mark · locally synthesized. On a small screen, use fullscreen for diagram details or read the transcript.
Read the descriptive transcript
0:00 · What you will learn
Roll out changes with an owner, a stop rule and a recovery plan.
What appears on screen
Roll out changes with an owner, a stop rule and a recovery plan.
0:07 · The idea
A new tool or data path changes what the workflow can do. Review that change before use. Keep a safe fallback and require evidence before restarting after a failure.
What appears on screen
A new tool or data path changes what the workflow can do. Review that change before use. Keep a safe fallback and require evidence before restarting after a failure.
0:20 · One useful term
Rollback. Returning to a previous version; the restored behavior must still be checked.
What appears on screen
Returning to a previous version; the restored behavior must still be checked.
0:29 · A worked example · 1
Version one only drafts. Version two adds sending and drops source IDs. When an unapproved send is attempted, pause the workflow and follow the established incident process.
What appears on screen
Version one only drafts. Version two adds sending and drops source IDs. When an unapproved send is attempted, pause the workflow and follow the established incident process.
0:44 · A worked example · 2
Use manual drafting while it is repaired.
What appears on screen
Use manual drafting while it is repaired.
0:50 · Your turn to think
Does returning to the old version automatically approve restarting? Pause and decide before continuing.
What appears on screen
Does returning to the old version automatically approve restarting? Pause the video if you need more time.
1:07 · Why this answer works
No. Check the restored behavior and get the required decision from the responsible owner.
What appears on screen
No. Check the restored behavior and get the required decision from the responsible owner.
1:16 · Check 1
Name who can stop the workflow.
What appears on screen
Name who can stop the workflow.
1:20 · Check 2
Preserve evidence under the handling rules.
What appears on screen
Preserve evidence under the handling rules.
1:26 · Check 3
Check recovery before resuming.
What appears on screen
Check recovery before resuming.
1:31 · Now try a different case
Open the lesson and try its different practice case. Use every check in the full task. Watching alone does not mark the lesson complete.
What appears on screen
Open the lesson Try the new case Check and revise The full task and its requirements are in the linked lesson.
Watching explains the example. Complete the lesson practice to check what you can do. Explore all visual lessons →
Before you start
Prerequisite: A6 gate decision and A2 versioned package. Offline/no-cost route: paper, calculator and the synthetic facts below. Optional tools must already be authorized. No paid API, private workplace record, account creation or external action is required.
What you will be able to do
- Write an SOP with roles, gates, fallback, and escalation.
- Classify changes and regression triggers.
- Contain, report, recover, and learn from incident.
Part 1 Changes and incidents need different controls
An SOP states intake scope, operator steps, review gates, exceptions, fallback, records and owner. Track changes to prompt, model/provider, data, retrieval, tool permissions, schema and reviewer process. A formatting edit needs proportionate checks; added write access or a new data source changes risk and authorization. Incident flow begins with detection and containment, then authorized evidence preservation/reporting, recovery, communication and learning. Do not delete suspected spill evidence or spread it by forwarding. Restoration alone does not demonstrate safe recovery; repeat relevant regression cases and obtain required authority before resuming.
Part 2 Worked synthetic example
Synthetic v1.0 is draft-only and passes schema tests. v1.1 adds a send tool and drops source IDs. Operator notices an unapproved send attempt at 10:05. Pause this workflow, disable its action path under established scope, preserve synthetic logs with access controls and report to fictional owner Lee. Use manual drafting while scope and provenance are repaired. A rollback to v1.0 restores draft-only behavior, but must be checked and authorized before operational resumption. No real incident report is sent in this tabletop.
Part 3 Guided practice and feedback
Classify changes: typo in heading -> rendering check; new model -> quality/regression checks; new connector/write permission -> authorization plus security/behavior review; retention change -> records/data authority. If sensitive content appears, report through policy rather than trying to erase traces. Record uncertainty when you do not know whether spill occurred.
Review and next step
Self-review is learning evidence, not independent certification or operational approval. A8 combines scope, evaluation, handoff and change evidence in a capstone.
Substantive review: 2026-10-02. See the configuration guide and the NIST AI RMF, a voluntary framework rather than a course authorization or credential.