The first 180 days as an AI leader
Use checkpoints to decide when scaling is justified.
In this lesson · 07 of 7
All lessons in this track →Saved in this browser. View your learning.
Start with the idea
Use checkpoints to decide when scaling is justified.
A plan should produce evidence at each checkpoint. Check quality and operations before expanding. A new model is another configuration to test, not an automatic upgrade.
Conditional hold: Continuing only the permitted limited work while a required gap remains unresolved.
See the example and check your understanding
A made-up teaching example
At day ninety, NORTHSTAR passes source checks but cannot restore a corrupted index. The team holds operational expansion, continues the offline exercise and assigns a recovery-test owner.
Ask yourself
Do strong answers justify scaling without a recovery path?
Show the explanation
No. Keep the hold until the required recovery evidence exists and the responsible decision is made.
Start with these checks
- Require quality and operational evidence.
- Assign unresolved risks to owners.
- Revise specific tests when assumptions change.
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.
The first 180 days as an AI leader
Use checkpoints to decide when scaling is justified.
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
Use checkpoints to decide when scaling is justified.
What appears on screen
Use checkpoints to decide when scaling is justified.
0:06 · The idea
A plan should produce evidence at each checkpoint. Check quality and operations before expanding. A new model is another configuration to test, not an automatic upgrade.
What appears on screen
A plan should produce evidence at each checkpoint. Check quality and operations before expanding. A new model is another configuration to test, not an automatic upgrade.
0:20 · One useful term
Conditional hold. Continuing only the permitted limited work while a required gap remains unresolved.
What appears on screen
Continuing only the permitted limited work while a required gap remains unresolved.
0:30 · A worked example
At day ninety, NORTHSTAR passes source checks but cannot restore a corrupted index. The team holds operational expansion, continues the offline exercise and assigns a recovery-test owner.
What appears on screen
At day ninety, NORTHSTAR passes source checks but cannot restore a corrupted index. The team holds operational expansion, continues the offline exercise and assigns a recovery-test owner.
0:45 · Your turn to think
Do strong answers justify scaling without a recovery path? Pause and decide before continuing.
What appears on screen
Do strong answers justify scaling without a recovery path? Pause the video if you need more time.
1:03 · Why this answer works
No. Keep the hold until the required recovery evidence exists and the responsible decision is made.
What appears on screen
No. Keep the hold until the required recovery evidence exists and the responsible decision is made.
1:12 · Check 1
Require quality and operational evidence.
What appears on screen
Require quality and operational evidence.
1:18 · Check 2
Assign unresolved risks to owners.
What appears on screen
Assign unresolved risks to owners.
1:23 · Check 3
Revise specific tests when assumptions change.
What appears on screen
Revise specific tests when assumptions change.
1:29 · 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 →
Sequence evidence before scale. Build a judgment you can explain and a result another reader can inspect.
Prerequisites: know the difference between generated text and source evidence. For configuration vocabulary, use the AI workflow reference. This is public self-study, not official government training, accreditation, a clearance or a professional credential.
Observable outcomes: Build an evidence-gated 180-day plan; reopen specific tests when network, model or reviewer assumptions change.
Part 1 Sequence evidence before scale
A 180-day plan sequences learning and legitimate decisions. Days 1–30 establish mission, baseline, boundaries and owners. Days 31–60 build a bounded synthetic or authorized pilot and test failures. Days 61–90 review performance, operating capacity and risks. Days 91–180 scale only where acceptance evidence and responsible approvals support it. Dates are checkpoints, not promises that permission will arrive.
Include one non-AI alternative, one stop decision and an exit path. Changes to model, retrieval, tools, permissions, hosting or policy reopen the relevant evidence. Use weekly operational reviews, periodic risk checks and a tested incident exercise sized to the mission; meeting frequency is not itself governance quality. Keep a decision log with the evidence, owner, conditions and trigger for reconsideration.
A useful leadership brief answers what changed, who benefits, what failed, which uncertainty remains and what decision is requested. Protect the team from pressure to strengthen claims without evidence. Practice on the fictional NORTHSTAR service; no live program documents, external messages or deployment are required.
Part 2 Synthetic worked example
At day 90 NORTHSTAR passes source-support checks but cannot restore its index after corruption. The team records a conditional hold on operational scale, continues a reversible offline exercise and assigns a recovery test owner. When a new model becomes available, it is a new candidate configuration to test, not an automatic upgrade.
Guided practice: write the tempting conclusion first, then underline the fact that does not support it. Rewrite the conclusion with that limitation visible. Identify one additional observation that would change the decision.
Next step: carry your artifact into the AI workflow reference; retain assumptions and unanswered questions.