Prompt Engineering for Public Speaking: How to Get Better Results from AI
Liam Sandford
Liam Sandford is a Head of Marketing, public speaking expert, and 2x Best Selling Author including the book Effortless Public Speaking. He helps ambitious professionals and entrepreneurs communicate with impact to get noticed, grow their career, and build their business.
Most people try AI for the first time, type something vague, get something generic back, and decide the tool is overhyped. The problem is almost never the AI. It is the prompt.
The gap between a forgettable AI output and one that genuinely sharpens your preparation comes down to how clearly you say what you need. That is not a developer's skill. It is the same skill that makes you better on stage: clarity, specificity, and knowing the outcome you want before you open your mouth. If you can do that in front of a room, you can do it in a prompt.
If you have been poking at tools like Chat GPT or Claude to prepare a presentation and finding the results flat, the fix is simpler than you think. You need to get better at asking.
Why Vague Prompts Produce Vague Speeches
"Help me write a speech about leadership" gives AI nothing to work with. It does not know your audience, your context, your experience, or what you want people to do afterwards, so it fills the gaps with the most average leadership speech it can assemble. The output is generic because the input was.
This is the same rule that governs speaking itself. If your core message is vague, your delivery will be vague, because clear beats clever and a confused speaker makes a confused room. The quality of your input decides the quality of the output, on the page and on the stage. The more specific you are about who you are speaking to, what they care about, and what you want to achieve, the more useful every response becomes.
How to Structure a Prompt That Gets Useful Results
A strong prompt for speaking preparation has four parts: context, audience, task, and constraints. It is the same brief you would give a sharp colleague.
Context sets the situation: "I am giving a 15 minute presentation at a regional sales conference for financial advisers who are sceptical about adopting new technology."
Audience says who they are and what they value: "They are experienced professionals who care about client relationships more than automation."
Task says exactly what you need: "Give me three opening hook options that acknowledge their scepticism while creating curiosity about the upside."
Constraints set the boundaries: "Keep each under 40 words, avoid jargon, and use conversational language I can deliver naturally."
Provide all four and the output stops being generic advice and starts being material you can work with. One thing I do that lifts the quality further: I have my own brand guidelines and a few saved workflows loaded into the tool, so the output already sounds closer to me than a cold prompt would. You do not need that to start, but it shows where this goes, the more context the tool holds about you, the less editing you do.
Why Your First Prompt Should Never Be Your Last
The most common mistake is treating AI like a search engine: type one query, take the first result, move on. That throws away most of its value.
The real power is in the follow up. After the first response, push back. Ask it to challenge its own suggestion, request alternatives, or tell it exactly what felt off and why. The single most useful follow up I have is the brutally honest pass: "act as a sceptical member of my audience, tell me where this loses you and which claim you do not believe." It stings the first time, then it becomes the prompt you reach for most.
This is how you would work with a coach or a trusted colleague. You would never accept the first draft of an opening without testing and refining it. AI works the same way, but only if you treat it as a conversation rather than a vending machine.
How Prompt Specificity Changes Every Stage of Speech Preparation
Prompt engineering is not about one perfect prompt, it is about asking the right question at each stage.
When you are exploring ideas, broad prompts work: "What are the biggest misconceptions about [topic] among [audience]?" gives you raw material to react to.
When you are structuring, precision matters more: "Using the Nano Speech framework of open, body, close, give me three structural options for a presentation on [topic] aimed at [audience] where the goal is [outcome]."
When you are refining the close, tighten the prompt further: "Rewrite this closing paragraph to be more direct, under 30 words, and end with a clear action the audience can take today." If you can say it in 5 words, do not let it sprawl to 10. Each stage needs a different kind of prompt, and knowing which to use when is what separates useful AI from frustrating trial and error.
How to Use Prompts to Stress Test Your Core Message
One of the most valuable uses of prompting is pressure testing what you plan to say before you say it. Most speakers skip this and only find the weak points when they are on stage, when it is too late.
Ask AI to poke holes in your argument: "Here is my core message: [message]. What would a sceptical audience member challenge about this, and where is my reasoning weakest?" That forces you to face the gaps you have been stepping around.
You can also test how the message lands with different rooms: "How would a room of CEOs read this versus a room of early career professionals?" The answers help you adjust your framing without changing your substance. All of this is one stage of the wider AI workflow for public speaking.
Why Role Based Prompts Produce Better Feedback
One of the most underused techniques is asking AI to adopt a specific perspective before it answers. Instead of "give me feedback on this opening," try "act as a sceptical audience member who has sat through dozens of presentations on this topic, what is your honest reaction?" Forcing the response through a defined lens produces sharper, more targeted feedback than a generic review.
Use it to:
Simulate a hostile Q&A panel so you can rehearse the tough questions
Get the reaction of someone who knows nothing about your topic
Test whether your close works for the decision makers in the room rather than the influencers
The key is to define the role clearly. The more specific the persona, the more specific the feedback.
How to Avoid Over Engineering Your Prompts
There is a tipping point where a prompt becomes so elaborate it defeats the purpose. If you spend longer crafting the perfect prompt than you would spend just thinking the problem through yourself, you have overcomplicated it.
A good prompt is specific, not exhaustive. It gives AI enough to be useful without turning every request into a briefing document. The test is simple: could you explain what you need to a knowledgeable colleague in under 30 seconds? If yes, that is about the right level of detail. If you are writing a page of instructions, break the task into smaller pieces instead.
How to Build a Prompt Library for Recurring Speaking Situations
If you speak regularly, you keep hitting the same preparation challenges, so a small library of proven prompts saves time and keeps your output consistent. Keep the prompts that have genuinely worked and organise them by stage:
Ideation prompts for fresh angles on familiar topics
Structure prompts for testing different frameworks
Refinement prompts for tightening language and cutting filler
Rehearsal prompts for simulating audience questions and objections
This is not about building a rigid system. It is about lowering the friction of starting, so you spend more time on the work that matters: understanding your material well enough to deliver it with confidence and flexibility.
What Prompt Engineering Cannot Replace
No prompt, however well crafted, replaces internalising your material. AI can help you think more clearly, structure more efficiently, and prepare more thoroughly. But the moment you stand up, what matters is how well you understand your content, not how polished the AI output was.
Prompt engineering is a preparation skill. It makes the preparation faster and sharper. It does not make you better in the room. That comes from reps, presence, and a willingness to connect with the people in front of you.
Frequently Asked Questions About Prompt Engineering for Public Speaking
What is the best prompt format for preparing a speech with AI?
The most reliable format has four parts: context (the situation), audience (who they are and what they care about), task (exactly what you want produced), and constraints (length, tone, and style boundaries). Provide all four consistently and the output is genuinely usable rather than the vague answer an open ended request returns.
How many prompts should I use to prepare one speech?
There is no fixed number, but useful preparation usually means several exchanges per section: start broad to explore, then narrow with follow ups that refine, challenge, and tighten. Treat it as a conversation rather than a single query, and stop when the material is clear enough to rehearse.
Can prompt engineering help with nerves before a presentation?
Indirectly, yes. Better prompts produce sharper preparation, and sharper preparation builds real familiarity with your material. That familiarity is one of the most dependable ways to calm nerves, because when you understand your content deeply you lean on it rather than on memory.
Should I use the same prompts for every speech?
Reuse the structure of prompts that have worked, but always change the context and audience. A library gives you a starting point, not a script to paste without thinking. The output is only ever as useful as the specificity behind it, and that has to be rebuilt for every new room.
TL;DR: Prompt Engineering for Public Speaking
Better prompts produce better AI output, which produces sharper preparation. The skill is the same one that makes you clear on stage.
Build every prompt from four parts: context, audience, task, and constraints.
Never stop at the first answer; follow up, challenge it, and run the sceptic's pass over your draft.
Match the prompt to the stage: broad for ideas, precise for structure, tight for the close.
Keep a small library of prompts that have worked, organised by preparation stage.
Remember no prompt replaces understanding your material well enough to deliver it without the notes.
More From Liam Sandford
Read my book: Effortless Public Speaking. Learn how to speak confidently, reduce stress, and turn public speaking into your competitive advantage. These actionable public speaking tips will help you improve your presentation skills for any audience.
Join the free 5-day email course: Get daily lessons packed with practical strategies to deliver effective presentations and speak confidently. This course is designed to build your public speaking skills step by step. Sign up below: