Talk With Legends: 17 ChatGPT Persona Prompts That Let You Converse With Icons
Some of the most useful thinking I've done with ChatGPT hasn't come from asking it to write something or analyze something. It's come from conversations where I assigned it a persona and then let that persona push back on my ideas.
I've had "Socrates" question the assumptions in my business strategy until I couldn't defend two of them. I've asked "Richard Feynman" to explain a technical concept to me using first principles until I actually understood it. I've put a rough creative brief in front of "David Ogilvy" and gotten the kind of sharp, no-nonsense feedback I'd be lucky to hear from a real creative director.
None of this is magic. It's a prompting technique — persona prompting — and once you understand what makes it work, it becomes one of the most flexible tools in your AI workflow.
How Persona Prompts Actually Work
When you assign ChatGPT a persona, you're doing something specific: you're telling the model to use the patterns, vocabulary, reasoning style, and worldview associated with that figure when generating its responses. The model draws on everything it learned about that person during training — their documented thinking, writing, public statements, methods, and philosophy.
The result isn't a simulation of the actual person. ChatGPT doesn't "become" Seneca. What it does is generate responses filtered through the lens of Stoic philosophy, the moral framework Seneca articulated in his letters, and the direct, unsentimental writing style that defines his work. For practical purposes, that's often more useful than talking to another generalist AI voice.
The technique works best for three things: getting a specific intellectual perspective on a problem, learning how a particular thinker's framework applies to your situation, and getting feedback filtered through a specific set of values and standards.
The Anatomy of a Good Persona Prompt
A weak persona prompt just names the person: "Respond as Einstein." That produces surface-level mimicry — some famous quotes, a vaguely Germanic syntax, maybe a thought experiment or two.
A strong persona prompt defines the context, the task, and the constraints:
Three additions that make persona prompts significantly more useful:
"Apply their thinking to my specific situation" — this prevents the model from producing generic biographical content about the persona instead of actually engaging with your problem.
"Be direct and willing to challenge my assumptions" — without this, ChatGPT tends toward agreement and validation regardless of the persona. This instruction overrides that tendency.
A specific question or problem to react to — open-ended conversations with personas tend to stay shallow. Give them something concrete: a decision you're trying to make, a piece of work you want feedback on, a concept you want to understand better.
17 Persona Prompts Worth Using
Here's my working list, organized by the type of thinking each persona is most useful for.
For strategic and business thinking:
Warren Buffett — on evaluating business decisions, competitive moats, and the discipline to say no to most opportunities. Ask Buffett to review your business model, pricing strategy, or investment in a new initiative.
Peter Drucker — on management, organizational effectiveness, and the right questions to ask about purpose. Useful when you're trying to decide what your business should stop doing.
Sun Tzu — on competitive positioning, knowing when to fight and when not to, and leveraging your opponent's weaknesses. More useful than it sounds for marketing and competitive strategy.
Charlie Munger — on mental models, thinking errors, and inversion (thinking about what you're trying to avoid, not just what you want). Excellent for decision-making reviews.
For creative and communication work:
David Ogilvy — on advertising, headlines, and the discipline of writing for results rather than admiration. Ask Ogilvy to critique your copy. He'll be blunt in ways that are useful.
Stephen King — on storytelling, tension, and the specific craft of writing that makes readers keep going. Useful for long-form content and narrative structure.
Ernest Hemingway — on cutting. Hemingway's persona will tell you everything in your writing that's doing nothing. Useful when your copy is too long and you can't see what to remove.
Bill Bernbach — on the difference between clever advertising and effective advertising, on emotional truth, on finding the one thing. For brand and campaign work.
For philosophical and ethical reasoning:
Socrates — for questioning your assumptions through a series of probing questions. Assign Socrates to interrogate any decision or belief you hold with high confidence. Plan to be uncomfortable.
Seneca — for Stoic perspective: distinguishing what's in your control from what isn't, thinking clearly about risk, and approaching difficulty without drama.
Marcus Aurelius — on duty, discipline, and how to behave well when circumstances are difficult. Useful for leadership and ethical decision-making.
For scientific and analytical thinking:
Richard Feynman — for first-principles explanation. Ask Feynman to explain any concept you're struggling with. He'll tell you immediately if the thing you're trying to understand is actually understood, or just described in jargon.
Charles Darwin — for patient, evidence-based reasoning and careful observation. Useful for research-oriented thinking and building conclusions from accumulated evidence.
Albert Einstein — for thought experiments, challenging assumptions about how systems work, and thinking about problems from unexpected angles.
For design and innovation:
Steve Jobs — for ruthless simplification, the user experience perspective, and the question "what does the customer actually experience?" Can be useful for product feedback. Also useful for identifying everything you've built that you shouldn't have.
Dieter Rams — for the "less but better" framework applied to design, products, and communication. Good for editing any deliverable where you're not sure what to cut.
Leonardo da Vinci — for cross-disciplinary thinking and the habit of observing before concluding. Useful for creative projects that benefit from an unusually wide frame of reference.
Getting the Most From These Conversations
The persona prompt opens the conversation. What you put in front of the persona determines how useful it gets.
Paste in your actual work. Don't describe it — show it. "Here's my landing page copy. Respond as David Ogilvy and tell me what's working and what isn't." The model produces sharper feedback when it has specific material to react to.
Ask for disagreement explicitly. "Where would you push back on this?" "What am I missing?" "What's the weakest part of this argument?" Personas default toward helpfulness just like regular ChatGPT. You have to prompt for honest critique.
That three-question structure prevents the persona from giving vague feedback. It forces specific, actionable responses.
Frequently Asked Questions
How accurate are these personas?
The accuracy depends on how extensively the person's thinking has been documented. Historical figures with large written bodies of work (Seneca, Marcus Aurelius, Darwin) tend to produce more reliable personas than people known primarily through media appearances or whose documented philosophy is limited. For famous business figures with significant documented interviews and writing (Buffett, Munger, Ogilvy, Drucker), the personas are quite useful. For figures where the historical record is sparse or contested, treat the outputs as inspired by rather than representative of the actual person's thinking.
Can I combine personas?
Yes, and it can produce interesting results. "You are David Ogilvy and Richard Feynman in conversation. Ogilvy is reviewing this marketing copy and Feynman is challenging his assumptions about what drives human decision-making." Multi-persona conversations introduce productive tension that single-persona conversations can't.
What if I want a persona that isn't a historical figure?
You can construct composite personas: "You are a venture capitalist with 20 years of experience who has backed 3 unicorns and passed on 5 more. You're known for asking the one question founders can't answer. Review my pitch." This approach lets you create specialist perspectives that don't correspond to any single person but draw on the archetype of that role.
Is there a risk of getting wrong information from a persona?
Yes. The model draws on training data, which may contain inaccuracies about any given person's documented views, or may attribute positions to them that they didn't hold. Use persona conversations for perspective and frameworks, not for accurate biographical or factual claims. If a persona quotes someone or cites a specific argument, verify it independently.
