Target the Mission
What outcome matters, and how will we know it worked?
BORROWED FROM · COMMANDER'S INTENT
YOU LEAVE WITH
A Target Statement: purpose, end state, evidence and limits, with an owner and a Decider
The TURNS Method
Every AI win is a series of good turns.
TURNS helps a team pick the right AI pilot, test it honestly, keep people in charge, and decide on evidence whether to start, continue or stop.
TTarget the MissionUUnderstand the HumansRRehearse with AINNavigate the RulesSSustain and Debrief
START WITH T. RETURN TO ANY TURN WHEN THE EVIDENCE CHANGES.
The five turns
One proven habit from each world
Each turn borrows a habit from UX research, television production or mission planning. It's a loop, not a checklist: interviews in U can send you back to T, and live results in S can send you back to any turn.
Understand the Humans
Who does this work, where does it hurt, and what would earn trust?
BORROWED FROM · UX RESEARCH
YOU LEAVE WITH
A workflow map, an evidence trail and a trust baseline
Rehearse with AI
Can a rough version on real approved cases pass a dress rehearsal?
BORROWED FROM · TV TABLE READS AND REHEARSALS
YOU LEAVE WITH
A locked script and an independently scored rehearsal
Navigate the Rules
Where must a human review, approve or override? What is off-limits?
BORROWED FROM · RULES OF ENGAGEMENT
YOU LEAVE WITH
Decision rights and tested controls
Sustain and Debrief
What did we learn, what is next, and how do we keep it going?
BORROWED FROM · AFTER-ACTION REVIEW AND POST-PRODUCTION
YOU LEAVE WITH
A debrief, a recorded gate decision and a review rhythm
INSIDE R · REHEARSE WITH AI
Protect the final test
A rehearsal only counts if the AI never saw the final cases. Teach, tune and evaluate with separate cases, and lock the script before the sealed cases are opened.
Three threads
What runs through every turn
Evidence, human responsibility and iteration aren't extra stages. They never happen at only one point, so they don't get letters.
Evidence
Write down what you know, how you know it, and what you don't know yet. Missing evidence is never a pass and never a zero.
The Director stays in the booth
AI is the crew. People make the calls, and they need four things to do it well:
- The source material
- Time to actually check
- Authority to reject, edit or pause
- A route to escalate and restart
Iteration
Every mission, map and rule is a current version. Changing your mind on new evidence is the method working, not failing.
Three decision gates
Evidence opens a gate. The calendar doesn't.
Keeping three separate decisions stops a promising rehearsal from being mistaken for proven value. Every call is written down, and a well-supported "stop" is a good result.
GATE 1
Greenlight
Is this problem worth investigating?
Calls: Greenlight call (proceed, revise, park)
Who decides: The problem owner. The selection stays provisional.
GATE 2
On Air
Should we authorize a limited live pilot?
Calls: On Air call (go, adjust, no-go)
Who decides: A named Decider, only after required controls are verified.
GATE 3
Renewal
Should we continue, adjust or expand?
Calls: Renewal call (continue, adjust, expand, stop)
Who decides: A named Decider, on live evidence. New scope needs its own evidence.
Inside every turn · the prompt companion
ENABLE: six steps to a better AI prompt
TURNS guides the work of adopting AI. ENABLE frames each request inside that work, then checks what comes back.
| STEP | WHAT YOU DECIDE | ASK YOURSELF |
|---|---|---|
| E | Establish the aim The purpose, the people served, and what success looks like | What will this help someone do? |
| N | Name the perspective The expertise or viewpoint that sharpens the answer | Whose judgment should shape this? |
| A | Anchor in context Background, sources, examples, and which facts are confirmed or assumed | What must the AI know? |
| B | Build the brief The deliverable, format, method, and checkpoints | What should I receive, and where should it pause? |
| L | Limit the scope Audience, tone, length, exclusions, privacy, and human approval | What must it respect, avoid, or leave to me? |
| E | Evaluate The checks, feedback, and revisions | What will I verify, and what should change? |
After Evaluate, take another turn: revise the prompt or the work, then run it again.
Match the effort to the cost of being wrong
Quick
The stakes are low and the result is easy to check.
Steps: Establish, Build, Limit
Example: Tightening a short email
Standard
Most everyday work.
Steps: All six
Example: A client update or a one-page brief
High-stakes
Others will act on the result, or it touches money, people or public claims.
Steps: All six, plus a named approver, checkpoints and independent verification
Example: A pilot plan, a policy summary, an analysis for a decision
IMAGE SLOT · REPLACE WITH FINAL COVER · 2:3, 1600 × 2400 PX
The book · coming soon · cover concept
The full method, with the stories behind it
The Director Stays in the Booth walks through every turn, thread and gate with lessons from television, UX and mission software. Join the waitlist for launch news and an early chapter.
You're on the waitlist.
Check to confirm. You'll hear first when The Director Stays in the Booth is out.
What TURNS is not
- Not a software method. No coding is required to run a first pilot.
- Not legal clearance. It helps you spot when you need qualified advice. It doesn't replace it.
- Not a promise of results. It promises an honest decision, not a particular outcome.