On this page
Fix answers that miss the mark with the repair formula and a handful of power moves, without starting over.
Key takeaways
- When an answer misses, don't just say "make it better." Use the repair formula:
That's not it because ___. Great would look like ___. Try again.- Lead with the pain and end with questions.
- Don't start over when one part is off. Keep what works and fix the broken part.
- You run AI. AI doesn't run you.
Do this now: Pick three or four power moves and try them on a real task today.
Class Notes
🤔 Why power moves
Your prompt was great. The chat still went sideways. That happens to everyone.
Power moves are the short phrases that steer the car back without starting over. Save them, copy them, and pick three or four to use every day until they're habit.
You run AI. AI doesn't run you.
- Lead with the pain. Tell AI what's broken in one sentence before you ask it to fix anything.
- End with questions. "Please ask me any clarifying questions before you start." This one line turns AI into a thinking partner.
🔧 The repair formula
When an answer misses, don't just say "make it better." Say this:
That's not it because ___. Great would look like ___. Try again.Every move below is a way to fill in one of those blanks.
🩺 What went wrong? Pick your move
| When the answer is… | Try this |
|---|---|
| Generic or bland | Cast the role vividly · Slow down |
| Heading the wrong way | Start with the end · Push back |
| Long, fluffy or unfocused | Give it real stakes · Sixth-grade reading level |
| Technically right but flat | Tell AI the feeling you want |
| Close, but one part is off | Fix only the broken part · What do you need from me? |
| Ignoring rules you gave earlier | The audit clause |
| Unclear because you're not sure what you want | Interview me · Brain dump out loud |
| A draft that needs honest feedback | Rip this apart · Borrow a brain |
| Exactly what you wanted | Capture the winning prompt |
🚫 Don'ts
- Don't give vague feedback. "Make it better" doesn't tell AI what to change. Use the repair formula.
- Don't accept an answer without checking it. A confident answer isn't the same as a correct one.
- Don't start over when one part is off. Keep what works and fix the broken part.
🏗️ Before you hit enter
🎭 Cast the role vividly
You are a world-class [role]. You've worked with [the kind of clients you serve]. I'd pay you $1,000 an hour for this.Try it when: you start any new chat. Specific beats generic every time.
🙋 Confess what you don't know
I've never done this before. Treat me like a sixth grader and walk me through it step by step.Try it when: you're new to something. AI slows down and teaches.
🏁 Start with the end
Before we start, what are we really trying to achieve here? Tell me in plain English, and give me three checks for a great result.Try it when: you're about to start something big. Mid-chat version: "What are you assuming we're trying to achieve?"
⏰ Give it real stakes
This goes to [who] by [when]. What matters most is [the one thing]. If you have to choose, pick [clarity over length].Try it when: answers feel long, fluffy or unfocused. Stakes tell AI what to prioritize, and real stakes work better than made-up ones because AI can act on them.
💪 Better results from the start
🗣️ Brain dump out loud
I'm going to brain dump everything on my mind. Don't respond until I say I'm done.Try it when: you know something's wrong but can't say it cleanly yet. Talk it out with Wispr Flow, type it, or snap a photo of your handwritten notes.
🗂️ Let AI organize your head
I'm completely overwhelmed. Organize my brain dump into what matters most, what can wait, and what I'm missing.Try it when: you have 47 things swirling. Or try: "Ask me simple questions one at a time to pull this out of me."
❤️ Tell AI the feeling you want
I want the reader to feel like, wow, I should print this out and keep it.Try it when: results are technically right but flat. Describe the emotional outcome and let AI find the words.
Not sure what feeling you want? Ask: "What feelings could my reader have? Give me options."
The baby shower story
A friend asked ChatGPT for baby shower trivia about music, travel and babies. One question was about how many babies were stranded during a volcanic eruption in Iceland. Nobody knew any answers.
She never told AI the feeling she wanted: "I want everyone laughing, knowing most of the answers, and saying this was the most fun shower ever." Without the feeling, AI focused on the categories and missed the point.
🧑🏫 Borrow a brain
Can you be [expert] and give me feedback on this?Try it when: you want a fresh lens. Switch mid-chat: "Good. Now what would [another expert] say?"
Pro move: "Tell me who the best thought leaders are on [topic]. Then have each one give me feedback, one by one."
You get AI's take on that person's public ideas, not their actual advice. It's a lens, not an endorsement.
🔥 Rip this apart
Rip this apart as if you were [expert]. Tell me everything that's wrong with it and why it would fail. Then fix it using all those recommendations.Try it when: you have a draft (proposal, sales page, email, deck) and need honesty before it goes live. Don't ask "Is this good?" AI will be polite.
Score version: "Be the toughest judge at the Olympics. Score this 0–100 on how likely it is to [close the deal]."
🆘 When it goes sideways
Don't start over. Recover.
🐢 Slow down
Slow down and ask me 3 more high-impact questions so we get on the same page.Try it when: AI is heading the wrong way entirely.
🎤 Interview me
Interview me to extract what I actually need. Ask one question at a time.Try it when: you don't know what you want yet. Example: "Interview me with 25 questions to describe my voice, tone and brand."
❓ What do you need from me?
What information would you need from me to do this perfectly? And what are you assuming to be true right now?Try it when: the answer is close but not quite right. Flip it and let AI tell you what's missing.
📞 Who should I talk to?
Which thought leaders or experts would be great to consult on this?Try it when: you're in new territory. Level up: paste in the transcript of that expert's YouTube video or podcast so AI works from their actual words.
✋ Push back
Why do you think we're not aligning? What do you think I'm frustrated about? What would you need to nail this?Try it when: you keep going back and forth. Make AI explain its thinking, then correct it. Some limits are real (no access, a policy line), but many "I can't" answers just need more context.
🩹 Fix only the broken part
Keep everything that's working. Change only [the specific issue]. Tell me briefly what you changed.Try it when: one part is wrong and the rest is good.
🛠️ Keep what works
📈 How would you improve this prompt?
How would you improve this prompt so it always [includes the booking link and phone number]?Try it when: results are okay but not nailed. AI rewrites its own instructions.
📸 Capture the winning prompt
What prompt would reliably produce this result every time? Write it for me.Try it when: you just got exactly what you wanted. Save it so you can repeat it next month.
📊 Real-time critique
Be a [world-class adult learning expert]. Rate this 0–10 and critique it. What would make it a 10?Try it when: you're about to teach, present or publish something.
🔍 The audit clause
Before I accept this, audit your response against my requirements: [1], [2], [3]. Where did you nail it? Where did you miss? Fix anything that's off.Try it when: you gave rules early in a long chat. AI can lose track of early instructions as a chat gets long. Keep it to 1–3 requirements. The more you add, the weaker the check.
Example: a sales comp plan
You asked AI for a new comp plan with three rules: accelerators start at 100% of quota and cap at 150%, a team bonus only when the whole pod hits 90%, and simple enough to explain to a new rep in 2 minutes. Before taking it to leadership, run the audit clause against those three rules.
Simpler version: audit a follow-up email sequence so every email is under 100 words, includes a case study and ends with a soft call to action.
📖 Sixth-grade reading level
Deliver this at a sixth-grade reading level.Try it when: the output is dense or full of jargon. The test: could a 12-year-old follow it without re-reading?
🧠 Optional: advanced moves for high-stakes thinking
Optional. For big projects and decisions. Try these once the basics feel natural.
👥 Shadow Board
Build me a Shadow Board of 5 real thought leaders, authors, or experts for this project. Choose people with different lenses: strategy, psychology, execution, creativity, communication, and risk. For each person, tell me why they belong, what lens they bring, and what they will challenge. Use this board throughout the project to critique, pressure-test, and improve the work.💯 Board Scoring
Have each Shadow Board member score this from 0–100. For each person, show the score, what works, what is weak, and what would make it a 100. Then make the board debate, challenge each other, and create one final version designed to score 100/100 across all perspectives.🔮 Future Failure Audit
Imagine this failed 6 months from now. Tell me the 10 most likely reasons. For each one: the early warning sign, why we missed it, and how to prevent it now. Then tell me which failure is most likely to blindside us because it feels harmless today.✂️ Executive Anti-AI Polish
Give this an Executive Anti-AI Polish Pass. Make it sound like a sharp human operator wrote it, not AI. Cut anything repetitive, overexplained, generic, padded, performative, vague, too long, or trying too hard. Remove AI tells: repeated sentence patterns, filler transitions, corporate jargon, vague claims, motivational-poster energy, and anything that sounds written instead of said. Cut at least 20%. Deliver only when it is clear, human, sharp, concise, and impossible to mistake for AI.🚧 Invisible Resistance Scan
Scan this for invisible resistance. Where would a real human pause, doubt, lose trust, feel confused, feel pressured, need more proof, or stop moving forward? Then remove the friction and make the next step feel obvious, safe, and natural.🧱 Limiting Belief Detector
Scan this decision for limiting beliefs. Tell me where I may be operating from fear, scarcity, people-pleasing, old identity, avoidance, desire for certainty, or fear of being seen. For each one, tell me what belief is driving it, what it is costing me, and what a more powerful belief would be. Then help me make the decision from expansion, truth, and self-trust.⚖️ Universal Laws Check
Pass this situation through 5 universal laws: Alignment (does this match who I am becoming?), Expansion (does this move me toward growth?), Integrity (is this clean, true, and honest?), Energy (does this create life force or drain it?), and Reciprocity (does this create value for everyone involved?). Score each from 0–10. For anything below 10, tell me what is out of alignment, what needs to change, and what decision would score a 10.🎯 Your turn
Grab a prompt you've already written (the one from Building Great Prompts works great).
- Run it and find one weak spot in the answer.
- Choose three or four moves to try. Use the "What went wrong?" table to pick the ones that match.
- Compare the before and after. Save the winning prompt.
✅ Before you close this
📎 Resources
🌱 New to this?
Start with one move. The next time an answer misses, type: "Slow down and ask me 3 questions." That's it.
Stuck? Copy this page into Claude and say: "I'm new to this. Walk me through it one step at a time."
⬅️ Back to your pre-watch list: Helping Hands (all videos) · General · Advanced
