How Designers Are Actually Using AI in Client Work (Real Workflows, Not Theory)

---

How Designers Are Actually Using AI

Somewhere in the last eighteen months, I stopped treating AI like a novelty and started treating it like a member of my team.

Not the member who runs the strategy meeting. Not the one who gets on a discovery call or tells a client her logo direction isn’t working. But the one who handles the parts of a project that used to take me three times as long for no particularly good reason.

That shift changed everything about how our studio operates.

Designers are using AI most effectively in five places: research and competitive analysis before a project starts, first-draft copy the client then edits, site structure and page mapping, alt text and metadata at handoff, and post-launch SEO and AI-search audits. The common thread is that AI accelerates the parts of a project that aren’t the design. The thinking, the strategy, the taste, the client relationship? Still entirely you.

Here is exactly how designers are actually using AI in practice, because I think the design industry has too much vague conversation about AI and not enough “here is the actual prompt I used on Tuesday.”

Where Does AI Fit in a Real Design Project?

Let me walk you through a project timeline and show you where AI actually lands, because it’s not everywhere and it’s not nowhere. It’s specific.

Before the project starts: research and competitive analysis.
Before I open a single canvas, I use AI to do a first pass on a client’s competitive landscape. I’ll ask Claude or ChatGPT to analyze three to five competitors in the client’s niche and identify patterns in how they’re positioning, what they’re leading with, and where the gaps might be. This used to take me two to three hours of manual research. It now takes twenty minutes, and the output gives me a genuinely useful starting point for the strategy conversation.

A prompt I actually use: “Here are five websites from [client’s niche]. Analyze how each one positions their offer, what they lead with above the fold, and where there appear to be gaps in how they’re differentiating. Summarize in plain language.”

At discovery: site structure and page mapping.
After a discovery call, I use AI to help me draft an initial site map based on the client’s goals. I’ll feed in my notes from the call and ask for a recommended page structure with a brief rationale for each page. I don’t use the output verbatim. I use it to pressure-test my own thinking and catch anything I might have missed.

During the build: first-draft copy.
This is the one that has saved the most time, and also the one that requires the most care. I use AI to generate first-draft copy for each section of a site based on information gathered in the discovery process. The client then edits it in her own voice. What she gets is a document full of words she can react to rather than a blank page she has to fill.

The key thing I always tell clients: the copy AI generates is a starting point, not a finished product. It captures the structure and the key messages. Her voice, her specific stories, her language, those have to come from her. AI can build the scaffolding. She has to move in.

At handoff: alt text and metadata.
This one is unglamorous and I love it. Writing alt text for every image and meta descriptions for every page used to be one of those tasks that took way too long and got done less carefully than it deserved because it came at the end of a project when everyone was tired. AI handles a first pass on all of it now. I review and refine, but the blank page problem is gone.

Post-launch: SEO and AI-search audits.
After a site is live, I use tools like RankMath alongside AI to audit how the site is performing in traditional search, and increasingly, how it’s being cited in AI-powered search tools. This is newer territory and it’s evolving fast, but checking whether a client’s site is getting picked up by AI search is becoming a real part of the post-launch conversation.

What Are the Best AI Tools for Designers Right Now?

I want to be specific here because “use AI tools” is not actually useful advice.

Claude. This is my primary writing and thinking tool. I use it for copy drafts, strategy documents, competitive analysis, and client communication templates. The quality of the output is consistently high and it handles nuance better than most tools I’ve tried.

ChatGPT. I use this alongside Claude, often for a second perspective on a direction or when I want to see a different approach to the same brief. Having two tools with different tendencies is genuinely useful.

Perplexity. For research that needs to be current. Perplexity searches the live web and cites its sources, which makes it far more reliable than asking a language model for up-to-date industry information.

Otterly. This one is newer to our workflow and I want to name it specifically because it’s doing something important: tracking whether your clients are being mentioned in AI-powered search results. As AI search becomes a bigger part of how people find businesses, this kind of audit is going to matter more. We cover it in depth inside DesignEdit.

RankMath. Not strictly an AI tool but it integrates AI features into its SEO recommendations in ways that have genuinely improved how efficiently I get a site ready for search at launch. We use and recommend it here.

We also have a dedicated post on AI tools for Showit web designers that goes deeper on the specific tools we’ve tested and actually kept in our workflow. Worth bookmarking.

Should I Tell Clients I Use AI?

Yes. And here is the framing that keeps it a strength rather than a discount request.

The conversation I have with clients sounds something like this: “Part of what allows me to deliver this project efficiently and thoroughly is that I use AI to handle the research, first drafts, and metadata, which frees me up to spend my time on the parts that actually require my judgment and expertise. You’re getting more strategic thinking, not less, because I’m not spending three hours manually writing alt text.”

Most clients respond really well to this. The ones who push back and ask for a lower price because “AI did part of the work” are usually clients who were already focused primarily on the output rather than the outcome. And that is useful information to have early.

The frame is important: AI is handling the parts that don’t require you. You are handling the parts that do. Those are not equivalent, and your pricing should not treat them as if they are.

Can I Use AI to Write Client Copy?

Yes, with a clear process and honest expectations.

AI-generated copy is a first draft, not a finished product. It can capture structure, key messages, and a general tone. It cannot capture a specific person’s voice, her stories, her particular way of seeing her clients’ problems, or the specific language that makes her brand feel unmistakably hers.

What I’ve found works well is using AI to generate a section-by-section draft based on discovery call notes, then sending it to the client with a clear brief: “These are starting points. Your job is to edit each section until it sounds like you. Add specifics, change words that feel off, and tell me anywhere it feels generic.”

That process moves much faster than starting from scratch, and it gives the client something to react to rather than a blank page to fill. We go much deeper on this in our post on how designers can use AI for client copy, which is worth reading alongside this one.

What Should I Never Hand to AI?

This is the part I want you to really hold onto.

Strategy calls. The discovery call is irreplaceable. The nuance of what a client says, how she says it, what she hesitates on, what she lights up about, none of that gets captured in a transcript accurately enough to hand to AI and get something useful back. Be present for that call.

The final positioning decision. AI can generate ten positioning options. It cannot tell you which one is right for this specific client, this specific market, and this specific moment in her business. That decision requires judgment that is built from experience, and it is yours to make.

Anything you cannot personally defend in a meeting. If AI generated a direction and you haven’t fully thought through why it’s right, you are going to struggle when a client pushes back. Only present work you can explain and stand behind, regardless of where the first draft came from.

Client data. This one is practical and important. Do not paste client questionnaire responses, contracts, financial information, or any personally identifying details into a public AI tool. Use general descriptions and anonymize anything sensitive. This is a real confidentiality issue and it’s worth being explicit with yourself about where the line is.

How Do I Build an AI Workflow Instead of Just Using AI Randomly?

The shift that made the biggest difference for me was treating prompts like brushes.

A painter doesn’t pick up a new brush every time she paints. She has tools she knows, knows what they do, and reaches for the right one for the right job. Prompts work the same way. Once I found a prompt that reliably produced useful competitive analysis, I saved it. Once I found one that generated first-draft copy in a structure I liked, I saved that too. I now have a library of prompts I reach for by project phase rather than starting from scratch every time.

The practical version of this: create a simple document with your go-to prompts organized by project stage. Research prompts. Copy prompts. Metadata prompts. Client communication templates. Review it at the start of each project and pull what you need. It sounds simple because it is, and it saves a genuinely surprising amount of time.

DesignEdit is built around exactly this approach. Every week, one practical AI workflow written specifically for designers, with the actual prompt included. Not theory. The thing you open on a Tuesday and use before lunch. If you’re not subscribed, that’s where I’d start.

How Much Time Does This Actually Save?

In our studio, somewhere between six and ten hours per project depending on the scope.

The research phase alone went from two to three hours to about forty-five minutes. Copy drafts that used to take half a day now take an hour, because the blank page problem is gone and the client has something to react to immediately. Metadata and alt text at handoff went from ninety minutes to twenty.

That is not insignificant. That is the difference between taking on one more project per month or having one more Friday afternoon to myself. Both of those things matter.

The time savings aren’t the most important part though. The most important part is where that time goes. It goes back into the strategic thinking, the client relationships, the parts of the work that AI cannot do and that clients are genuinely paying for.

If you want to build out a real AI workflow for your design business, and go deeper than this post covers, Design Strategy School includes the systems and frameworks for running a modern design studio in the landscape that actually exists right now. Not the one from three years ago.

And if you want a weekly dose of this delivered to your inbox in a format you can actually use, DesignEdit is that. One workflow. One prompt. Every week.

---

FAQs