PLAYBOOKS
The Sam Altman Idea Evaluator
The custom prompt that runs your business idea through Sam Altman's Startup Playbook the way a YC partner would in a real interview. Free, ten minutes, brutally honest, full prompt included.
Read →
BY STEVE TAN
AI isn't a tool. It's leverage. Sharing what's working week by week.
PLAYBOOKS
The Operator's ChatGPT Prompt Book
100 prompts I actually use to run my businesses. Organized the way an operator's day actually breaks down: Think, Write, Explain, Create, Decide, Learn, Ship, Stack. Pick the workflow, grab the prompt, get to work.
Steve Tan
TL;DR
Most prompt lists are noise. People stack 100 prompts they've never used and dress them up like secret codes. This isn't that. 100 prompts I actually use, organized by what you're trying to get done, not what each prompt technically does. Eight workflows match the way an operator's day breaks down: Think, Write, Explain, Create, Decide, Learn, Ship, Stack. When you're stuck, you scan the workflow that matches what you need, grab the prompt, paste it into your chat. Power comes from stacking them. The model isn't the moat. How you use it is.
I get sent prompt lists every week. Most of them are noise. People stack 100 prompts they've never used, dress them up like secret codes, and sell the idea that AI is some hidden game with cheat codes.
That's not what this is.
The way I see it, AI is leverage. Same as relationships, same as capital, same as supply chains. The operators who win aren't the ones who know the most prompts. They're the ones who know which prompt to use when.
So I organized this book differently from the prompt lists you've seen before. Not by what the prompt does. By what you're trying to get done.
Think. Write. Explain. Create. Decide. Learn. Ship. Stack.
Eight workflows. That's how an operator's day actually breaks down. When you're stuck, you scan the workflow that matches what you need, grab the prompt, paste it into your chat. That's it.
The model isn't the moat. How you use it is.
Alright. Let's go.
THINK
Before you decide, before you write, before you build. The harder problems start here.
/firstprinciples — Strips an idea back to fundamentals. The way I think about most business problems. What's actually true here, before all the assumptions stacked on top?
/assumptions — Surfaces the hidden assumptions in a claim or plan. Most bad decisions trace back to assumptions nobody questioned.
HIDDENASSUMPTIONS — More aggressive version. Digs out the assumptions you didn't even know you were making.
META — Examines the framing of the question itself. Sometimes the question is wrong, and you don't need a better answer, you need a better question.
X10THINK — Forces the model to think 10x deeper before answering. Use it for the questions that actually matter.
DELTR — Decompose, Evaluate, Limit, Transform, Rebuild. Five steps for breaking a complex problem apart and putting it back together better.
FRACTAL — Zooms in and out across levels. Great for systems thinking. How does the strategy show up in the daily operation?
LENSSTACK — Same topic through multiple perspectives. Operator, customer, investor, competitor. The picture gets clearer when you stack lenses.
SYSTEMMAP — Maps the actors, forces, and feedback loops in a system. Use it before entering any new market or industry.
PARADOX — Surfaces the contradictions inside an idea. The tension is usually where the truth lives.
SIGNALVSNOISE — Separates what matters from what's distracting you. Most operators are drowning in noise. This is the filter.
HYPOTHESISGEN — Multiple explanations for the same observation. The first explanation is rarely the right one.
REVERSEENGINEER — Works backward from the outcome. How did this thing actually get built? What were the real moves?
BLACKSWAN — Rare but high-impact possibilities. The stuff 99% of people don't plan for, and the 1% that does runs the world.
/why — Forces the reasoning. Not just what something is, but why it works that way.
/deep — Pushes for layers. Don't give me the surface, give me what's underneath.
WRITE
When you've got something to say but it's not landing yet.
/human — Strips out the AI-sounding stuff. Makes it read like a real person wrote it. The way you'd actually say it to a friend.
/rewrite — Rephrases without changing the meaning. Useful when the idea is right but the words aren't.
/improve — Pushes the writing up a level. Better flow, clearer thinking, less fluff.
/shorten — Cuts the fat. Same message, half the words. Most things you write should be shorter. This forces it.
/expand — Adds depth, context, supporting detail. Use when something feels too thin to land.
/simplify — Drops the jargon. Plain language only. If your audience isn't in the trenches, run everything through this.
/formal — Tightens the tone for boardrooms, legal docs, investor updates. Less you, more polish.
/casual — Loosens it up. For DMs, captions, anything where stiff writing kills the connection.
/tonechange — Shifts the emotional or stylistic tone. Same words, different feel.
Create a free account to continue reading
Every Framework, Playbook,
and Prompt — Free, Forever.
The operator's library for building with AI.
“The most actionable AI resource library
I've found. Thanks Steve!”
James.H — Member since 2026
Join 2,845+ leaders, builders, and innovators
Already have an account?
/clarify — Tightens up ambiguous language. If you're not sure what something means, run it through this before sending.
/reframe — Same idea, different angle. Sometimes that's all it takes to unlock something.
/compress — Shrinks content while keeping the core. Different from /shorten. This preserves substance, not just word count.
EXPLAIN
Same idea, different altitudes. Pick the one that matches the room.
ELI5 — Explain it like I'm 5. For when something feels confusing and you want it stripped to the bone.
EL10 — Same thing but with a bit more meat on it. Good for early-stage learning.
EL15 — Teen-level. Solid baseline for explaining business or tech concepts to non-experts.
ELICollege — Steps it up. More rigor, more nuance, real arguments.
ELIPhD — Advanced and technical. Use when you're going deep with someone who already knows the basics.
/analogies — Forces the model to explain through comparisons. I lean on this constantly. Relationships, dating, gaming, building. Analogies make abstract ideas stick.
FEYNMAN — Forces the simplest possible explanation. If you can't explain it simply, you don't understand it.
/teachme — Frames the response as a lesson, not a generic answer.
/examples — Adds concrete examples to make an idea stick. Generic ideas die. Specific ones land.
/usecases — Lists practical situations where the idea applies. Good for moving from theory to action.
MISCONCEPTIONS — Lists what most people get wrong about a topic. Sometimes the fastest way to understand something is to know what NOT to think.
CREATE
When you need volume, not perfection. Quantity first, quality after.
IDEAS10 — Ten ideas, fast. Don't filter. Just get them out.
ALT3 — Three different versions of the same thing. Good for testing direction before going deep.
IDEAGEN — Wider spread. Pushes the model to explore directions you wouldn't have asked for.
/brainstorm100 — Big batch mode. When you want to go from zero to a hundred ideas fast.
SCAMPER — Substitute, Combine, Adapt, Modify, Put to other use, Eliminate, Reverse. Old-school creativity framework. It still works.
REMIX — Mash two concepts together. How most original ideas actually get made.
HOOKS10 — Ten hook options for a piece of content. I never publish the first hook I write. I always pull from a batch.
HOOK10 — Same idea, shorter form. For quick testing. Captions, opening lines, subject lines.
/viralhooks — Hooks built specifically for attention. Don't use this without taste. Viral without substance is a trap.
/improvehooks — Refines hooks to make them sharper. Most hooks die in the first three words. This catches that.
SCRIPT — Formats the output like a video or dialogue script. Useful for shorts, ads, content prep.
THREAD — Structures the response like a social thread. Hook, build, payoff.
/storymode — Shifts the answer into a narrative voice. Stories beat facts every time when you're trying to move people.
/simulate — Simulates a scenario or conversation. I use this for roleplaying difficult convos before I have them. Partner negotiations, hard feedback, investor calls.
/futurehistory — Describes the future as if it's already happened. Powerful for vision-setting and storytelling.
/expandideas — Takes a rough idea and develops it. Great for early brainstorm output that's not ready yet.
DECIDE
When you've got a call to make and you don't want to make it on vibes.
DECIDE — Builds a structured decision framework. For when "let me think about it" has gone on too long.
PROSCONS — Classic. Pros and cons, side by side. Sometimes the simplest tool is the right one.
SWOT — Strengths, Weaknesses, Opportunities, Threats. Use it on a business, a partnership, a launch. Any decision with multiple moving parts.
COMPARE — Side-by-side comparison of two or more options. Good for vendors, tools, deals.
PARETO — Finds the 20% of actions that drive 80% of the result. The whole game, basically.
RISKMAP — Identifies the real risks. Not the obvious ones. The ones you keep avoiding.
SECONDORDER — What happens after the first thing happens. Most people stop at the first effect. The money's in the second and third.
/predict — Projects likely outcomes from current conditions. Not prophecy. Just thinking ahead.
/counter — The strongest counterargument to your position. If you can't beat it, your position is weaker than you think.
/steelman — Builds the strongest possible version of the opposing argument. Different from /counter. This is the version even the other side wishes they could articulate.
/devilsadvocate — Argues against your idea on purpose. Better to get torn apart in private than in public.
/debate — Both sides, balanced. Useful when you're emotionally attached to one view.
KILLCRITIC — Blunt critique. No diplomacy, no padding. Just where it's weak.
REDTEAM — Aggressive critique. Use it when you're about to commit to something big.
/critique — Reviews the work and shows what needs improvement. Don't ask for praise. Ask for the fix.
/reverse — Starts from failure and works backward. What would have to go wrong for this to fail? Then don't do those things.
OODA — Observe, Orient, Decide, Act. Military decision loop. Underrated for fast-moving operators.
LEARN
When you're trying to actually learn something, not just consume it.
QUIZME — Generates quiz questions from the material. Use it on yourself, not on others.
FLASHCARDS — Turns content into prompt-and-answer pairs. Spaced repetition friendly.
TESTME — Asks you questions instead of feeding you answers. Forces you to think.
SOCRATIC — Teaches by asking questions. The way good mentors teach. They make you find the answer yourself.
/socratic — Same energy. Switches the whole response style into question-led teaching.
TIMELINE — Organizes things chronologically. Useful for understanding how a market, a company, or a technology actually evolved.
/coach — Frames the answer as guidance from a mentor. Useful when you want perspective, not just information.
/insights — Pulls out the patterns and observations that aren't obvious on the surface.
SHIP
How you want the answer to land on the page, in the codebase, in the inbox.
TLDR — The shortest version that still has signal. For when you're scanning, not reading.
BULLETS — Turns the response into bullet points. Easier to scan, easier to action.
TABLE — Side-by-side comparisons. Use this whenever you're weighing options.
CHECKLIST — Converts the answer into something you can actually execute against.
STEPS — Step-by-step format. For processes, SOPs, anything you want to repeat.
/stepbystep — Sequential process instead of a wall of text. Use it for SOPs and onboarding docs.
/outline — Structures the content into a clean outline. Great before you write the long version.
/summary — Concise summary of the main points. Different from TLDR. More complete, less compressed.
/keypoints — Pulls the takeaways. The 20% that matters.
/breakdown — Splits a complex topic into smaller pieces. The first move on anything overwhelming.
/convert — Converts content from one format to another. Article to script, script to thread, thread to email.
/translate — Translates into another language. I use this in Mandarin and English contexts daily.
DEBUG — Finds likely bugs in your code. Saves an hour of staring at a screen.
EXPLAINCODE — Explains what the code is doing in plain English. Useful when you're inheriting someone else's work or learning a new stack.
OPTIMIZE — Improves code for speed, cleanliness, or efficiency. Run it after you've got something that works.
/systemrewrite — Rewrites text into something more structured and systemized. Useful for docs, processes, frameworks.
STACK
Power moves. The ones that compound when you combine them.
/autoprompt — Generates a better prompt from a rough request. Meta-tool. Use it when you don't know how to ask the question yet.
/godmode — Pushes for the most assertive, comprehensive response. Use sparingly. And check the output. Confidence isn't the same as accuracy.
How to actually use this
Three ways to make these earn their keep.
- Stack them. The real power isn't in any single prompt. It's in combining them. Run
/firstprinciples, then/steelmanthe result, thenSECONDORDERit. That sequence on the same problem will get you somewhere most people won't. - Build your own. Once you see the pattern, you'll start writing prompts that are specific to your business. Your products, your customers, your edge. That's where the leverage compounds.
- Don't outsource your thinking. AI is leverage, not replacement. The prompts that work best are the ones where you bring the context, the judgment, and the taste. The model does the heavy lifting on volume and structure. You stay the operator.
Where I'm taking this next
I'm building a full operating system for entrepreneurs running businesses with AI. Not a prompt list. An actual stack. Agents, workflows, ops layer, the whole thing.
If that's interesting to you, follow along. I'm sharing the build in real time. What works, what doesn't, what I'm changing my mind about.
And if this prompt book saves you an hour this week, send it to one person who'd actually use it. That's the only way these things travel.
Appreciate you. Let's build.
Steve Tan
Builder · Operator · Advisor
20+ years building businesses the hard way across eCommerce, SaaS, agency, education, and supply chain. $200M+ in revenue. Now I help business owners turn AI into their unfair advantage.
More about Steve
