AI Prompts for Design Competitions and Client Presentations

TL;DR: Check the specific competition's rules on AI before you enter: they range from an outright ban on AI-generated submission imagery to a disclosure requirement, and assuming yours is fine is how entries get thrown out. Use AI prompts for research, concept exploration and narrative writing, not for generating the artwork you submit, and never let a client find out you used more AI than you told them.

Can you submit AI-generated images to a design competition?

Only if that specific competition's rules say so, and a growing number say no. This isn't a question this page can answer for you, and you should be suspicious of any page that claims it can. Competition rules are set by the organiser, published in the entry brief or FAQ, and they change between editions of the same competition.

What we can tell you is the shape of the landscape. Some competitions ban AI-generated imagery from the final submission outright, treating it the same as any other misrepresentation of authorship. Others allow AI involvement anywhere in the process as long as you disclose it, usually in a short statement attached to the entry. Others say nothing about AI at all, and silence is not the same as a green light; it usually means the rules were last updated before this was a live issue, not that anything goes. A handful explicitly welcome AI as a tool as long as the design decisions are the entrant's own.

If you've already skimmed the rules for an entry fee, a size limit and a deadline and stopped there, go back and read the eligibility and originality clauses in full. That's genuinely the whole first step, and it's not a formality.

What actually changes between one competition's rules and another's?

The variable that matters most is whether the rule is about the final artwork or about the process. A ban on AI-generated renders in the submitted boards is a narrower restriction than a ban on AI use anywhere in developing the concept, and conflating the two is how entrants either over-restrict themselves for no reason or under-restrict themselves into disqualification.

The second variable is disclosure. A competition that requires you to state where AI was used is not necessarily banning AI. It's asking you to be honest about it, which is a different, and lower, bar to clear. Treat any disclosure requirement as non-negotiable paperwork, not as optional context.

If a rules document is genuinely silent and you can't find an FAQ that addresses it, email the organiser and keep the reply. That costs you five minutes and removes the single biggest risk in this whole process: guessing wrong about somebody else's rulebook.

Where can AI prompts help before you touch the entry file?

This is where most of the real value sits, and it has nothing to do with generating a submittable image. Prompts genuinely help with:

• Precedent and case study research: pulling together comparable projects, typologies and approaches so you're not starting from a blank page
• Brief and programme analysis: turning a dense competition brief or client brief into a checklist of hard requirements versus soft preferences
• Concept exploration in words: testing two or three directions against the brief before you commit hours to drawing one of them
• Site and context synthesis: organising research about a site's history, constraints and surroundings into something you can actually reference while designing

None of this touches the artwork question at all. You can do every one of these things and still submit an entry that's entirely your own drawing, your own render, your own decisions, because the AI never generated anything that ended up on a board.

What does a concept-exploration prompt look like?

The trick is asking for words, not images, and being explicit about that boundary in the prompt itself. Here's a starting block:

You are a design tutor helping me think through concepts, not generate final artwork.

Project: [type, e.g. community library extension]
Site/context: [location, condition, key constraints]
Brief requirements: [paste the actual brief or competition programme]
A reference or precedent I keep coming back to: [1-2, in plain language]

Give me:
1. Three distinct conceptual directions in plain language, each in 2-3 sentences
2. For each direction: the strongest argument FOR it against this brief, and the strongest argument AGAINST it
3. One open question I should resolve before committing to a direction
4. A one-line "big idea" summary for each, written the way I'd say it out loud to a jury

Do not describe finished imagery, rendered materials or lighting conditions — this is
concept thinking in words, for me to develop and draw myself.

The last line is doing real work. Without it, most models default to describing a rendered scene, which quietly turns a thinking exercise back into an image brief. If you want the block that actually generates architectural imagery once you've decided that's compatible with your rules, our architecture prompt generator covers those eight slots separately, and our broader AI prompts for architects collection has more on the research side of this.

How do you turn research into a presentation narrative?

Juries and clients aren't scored on how much information you gathered. They're persuaded by a story: a problem stated plainly, one insight that shaped the response, and decisions that visibly answer that insight in an order a listener can follow without your help. Most design teams have the decisions already, buried in weeks of sketches and meeting notes, and what's missing is the fifteen minutes of restructuring that turns notes into a narrative.

This is a genuinely good use of AI, with one condition: feed it your actual decisions and ask it to restructure them, rather than asking it to invent the reasoning for you. The difference matters more than it sounds like it should, because a narrative built from decisions you didn't actually make is the one that collapses under a follow-up question from a juror or a client who's paying attention.

What does a presentation narrative prompt look like?

Turn my project notes into a presentation narrative, in my own voice.

Project: [name/type]
Audience: [who's in the room, what they care about — jury, client, both]
My actual design decisions, in my own words: [paste rough notes, however messy]
Time limit: [e.g. 5-minute pin-up, 20-minute client walkthrough]

Structure the narrative as:
1. The problem or opportunity, in one sentence a non-specialist would understand
2. The one insight that shaped the response — not a list of influences, the one that mattered
3. How the decisions in my notes above answer that insight, in an order a listener can follow
4. A closing line that restates the big idea, not a recap of features

Keep my actual reasoning throughout. Do not invent design decisions I didn't give you.

That last constraint again: don't invent reasoning. A prompt that lets the model fill gaps in your logic is a prompt that hands your jury a story you can't defend when they ask the obvious next question.

How do you prepare for jury or client questions?

The same restructuring exercise works in reverse: ask the AI to argue against you before a real person does. This is a form of persona prompting: putting the model in a specific adversarial role, rather than asking it to be generically helpful, surfaces weak spots in a narrative while you can still fix them.

Take the role of a skeptical juror (or client, depending on context) who has seen many
projects like this one and is not easily impressed.

Here's my presentation narrative: [paste the narrative from above]
Here's the actual brief or client requirements: [paste]

Ask me the five hardest questions you would ask about this project, focused on:
- Places where my narrative claims a benefit the design doesn't clearly deliver
- Assumptions I've stated as fact without evidence
- The single weakest link between my "insight" and my actual decisions

Be specific to this project. Don't ask generic architecture-critique questions.

Run this before the real thing, not instead of rehearsing out loud. A written answer to a hard question and a spoken one under time pressure are different skills, and only one of them gets tested in the room.

Do you have to tell a client you used AI?

Practically, yes, and the honest reason has nothing to do with etiquette. If a client presents your work internally, signs off on a design direction, or simply asks how you got somewhere so fast, and later learns AI did more of the thinking than they were told, the damage isn't confined to that one disclosure. It reaches backward into everything else you've told them, including the parts that were entirely yours. Trust doesn't partially fail.

There's no single disclosure law that covers every jurisdiction, every contract and every kind of AI involvement, so if you're genuinely unsure what your agreement or local requirements demand, that's a question for your own contract and, where it matters, a lawyer, not something a blog post can settle for you. What isn't ambiguous is the practical standard: what you present to a client as your design work should be your design work. Research assistance and note-taking are fine, and usually not worth a special conversation. The actual concept, the actual decisions, the thing they're paying for, has to be authored by the person or team they hired, not generated and lightly edited.

Agencies juggling this across several clients at once have their own version of this problem; our AI prompts for agencies and client deliverables piece covers the wider account-management side of it.

What does a disclosure-ready client note look like?

If your practice wants a standard way to describe AI involvement to clients, whether in a project binder, a proposal, or a stage sign-off, this drafts one honestly instead of either overstating or hiding it:

Draft a short, honest note for a client project file describing how AI was used at this
project stage.

Stages where AI genuinely helped: [e.g. precedent research, brief analysis, drafting this note]
Stages where it did NOT touch the design: [e.g. the concept, material decisions, the drawings]
Tone: plain, factual, no marketing language, no apology either

Write two to three sentences. State clearly what AI did and did not do. Do not use
hedging language that makes our AI use sound bigger or smaller than it actually was.

The output is two or three factual sentences you can drop straight into a project file, and it works precisely because it names both what AI touched and what it didn't.

What can't AI do for your entry or your deck?

It can't verify code compliance, structural feasibility, fire egress or accessible clearances, none of which show up in a compelling image regardless of how it was made. It can't know a specific competition's disclosure rules; only that competition's own documents can tell you that. It can't originate a concept you're prepared to defend under questioning, because a defensible concept has to already live in your own reasoning before you ask anything to help you say it more clearly. And it can't repair a relationship after a client learns you understated how much of "your" work was actually AI's.

What it's genuinely good at is compression: turning a folder of research, notes and half-formed decisions into something a jury or a client can follow in the time you're given. That's a real, useful, unglamorous job, and it's the one worth handing over.

None of this replaces reading the actual rules of the competition you're entering, or having an honest conversation with the client whose project it is. The prompts above just make sure that when you sit down to do the parts only you can do (the design, the reasoning, the story), you're not also burning hours on the parts a tool can help compress.

If you want a broader library of prompts for pitching design work in general, not just architecture, our proposals that win collection and our rendering and visualisation prompts, for when imagery is genuinely compatible with your rules, are the natural next stops.