Transparency has quietly become part of the design deliverable. It used to be enough to hand over a polished comp and a clean file structure. Now, somewhere between the kickoff call and the final invoice, a client is likely to ask a version of the same question: “Did you use AI for this?” How a designer answers that question — not whether AI was involved at all — is what determines whether the client’s trust in the work goes up or down.
1. Why clients are asking now
Five years ago, clients rarely asked how a mood board was made. Today they might have generated one themselves in Midjourney over the weekend, or asked ChatGPT to draft their own brand brief before the first meeting. That familiarity cuts both ways. It makes clients faster collaborators, but it also makes them more alert to the possibility that the polish they’re paying for was produced in seconds rather than earned over days.
Underneath the question is usually one of two anxieties. The first is financial: if a render came out of KeyShot’s AI-assisted lighting presets or a hero image started life as a Midjourney prompt, is the studio still charging for the hours a fully manual process would have taken? The second is legal and reputational: AI-generated imagery carries real, unresolved questions about copyright and originality, and a client shipping a campaign under their own name wants to know they’re not inheriting someone else’s liability. Neither concern is unreasonable. Both deserve a direct answer rather than a dodge.
2. The two wrong ways to answer
There are two failure modes, and they sit at opposite ends of the same problem: neither one accurately represents what actually happened in the studio.
Hiding it
Denying or omitting AI involvement feels safe in the moment but is a trust liability waiting to mature. If a client later notices a telltale Midjourney texture, an oddly generic ChatGPT-drafted line of copy, or simply asks a more technical follow-up question, the damage isn’t about the tool — it’s about having been misled.
Over-crediting it
The opposite mistake: framing the whole project as “AI-assisted” to sound current, when in reality AI touched one stage of a much longer process. This undersells the concept work, the client conversations, the revisions, and the taste that shaped the final output — and it invites the client to wonder why they’re paying design rates for what sounds like prompt-typing.
What both get wrong
Both responses treat “AI” as a single yes/no fact about the project, rather than describing which specific stage it touched and which stages it didn’t. That specificity is exactly what a client is actually trying to find out.
3. Frame it the way a photographer explains retouching
Photographers have been fielding a version of this question for two decades: “Is that retouched?” The ones who answer well don’t get defensive about Photoshop, and they don’t pretend the image came straight off the sensor. They explain, plainly, that the lighting and composition were the work of skill and judgment on set, and that retouching refined — rather than invented — the result. Architects do the same thing with rendering software: nobody assumes a firm is less skilled because the client presentation was built in a rendering engine instead of hand-painted.
Design should be explained the same way. Midjourney might generate the first fifty exploratory thumbnails for a concept direction, but choosing which three are worth developing, and knowing why they work for this brand and not another, is not something the tool did. Figma AI might remove a background or resize a component set in seconds, but deciding what the layout should communicate is not automatable in any meaningful sense. The honest, confident framing is: AI accelerated specific, nameable stages — ideation, rough variations, background cleanup, upscaling — while concept direction, taste, and final refinement remained entirely the designer’s.
Name the stage, not just the tool
The most convincing version of this explanation is specific. “We used Midjourney to generate early mood-board variations, then art-directed and hand-refined the three directions we brought to you” lands very differently than a vague “we used some AI tools.” Specificity signals control. Vagueness signals evasion — even when nothing was actually hidden.
💡 Pro tip — If AI touched a deliverable in a way the client couldn’t visually detect on their own — a KeyShot AI denoise pass, an upscaled export, a ChatGPT-drafted first pass of copy you then rewrote — say so anyway. Volunteering the detail before it’s asked builds more trust than a technically-true non-answer ever will.
4. Build the conversation into the process, not just the answer
The best time to explain AI usage is before anyone has to ask. Waiting for the question to come up mid-project puts the designer on the back foot, answering under mild suspicion instead of stating a fact. A short framework, repeated consistently across projects, removes that dynamic entirely.
At the proposal stage: Add one plain-language line to the scope document — e.g., “Early concept exploration may use AI tools such as Midjourney or ChatGPT for speed; all final direction, refinement, and production work is done by our team.”
At kickoff: Say it out loud, not just on paper. A verbal mention in the first meeting normalizes the topic so it’s not a surprise later.
Mid-project, if asked: Answer immediately and specifically — name the stage, name the tool, and pair it with the judgment work that surrounded it. Don’t pause, don’t qualify, don’t change the subject.
At delivery: If a specific asset relied more heavily on AI — an upscaled render, a generated texture — flag it in the handoff notes so there are no surprises for legal or licensing review later.
The recurring mistake worth naming on its own is defensiveness. A designer who stiffens, changes the subject, or answers with a question of their own (“why do you ask?”) signals that something is being protected. Clients read hesitation as concealment far more readily than they read a straight answer as a problem. A calm, specific, three-sentence explanation — tool, stage, judgment — closes the topic almost every time. A vague or nervous one keeps it open indefinitely.
Closing thoughts
AI tools are not going to become less embedded in design workflows, and clients are not going to become less aware of them. The studios that come out ahead won’t be the ones with the strictest no-AI policy or the loudest AI-forward marketing — they’ll be the ones who can describe their process plainly, stage by stage, without flinching. Treated that way, “did you use AI for this?” stops being an accusation to deflect and becomes what it actually is: a reasonable question with a straightforward, confidence-building answer.
Design Daily Life · Notes on design, daily