A good AI design workflow uses AI for the parts of the job that are slow and repetitive, like research, mood exploration, rough comps and cleanup, and keeps a person in charge of strategy, taste and the final work. AI speeds up the middle of the process. It shouldn't own the beginning or the end.
Here's how that actually plays out in my studio, step by step.
Step 1: Strategy (no AI decisions here)
Every project starts with questions. Who's the audience? Who are the competitors? What's the one thing people should remember? Where will this live: Instagram, a storefront, a shoe box, a stadium screen?
I'll sometimes use AI to help organize notes or summarize research I've gathered. But the answers come from talking to the client and knowing the market. If the strategy is generic, everything after it will be too.
If you've never written a brief, my design brief template walks through the questions I ask.
Step 2: Mood and exploration (AI is useful here)
This is where AI earns its spot. Before I commit to a direction, I want to see the extremes:
- What if it's darker and grittier?
- What if it leans retro?
- What if it's all chrome and 3D?
- What if it's flat, bold and graphic?
Generating rough studies lets me and the client react to something concrete in a fraction of the time it used to take to pull references. It's a sketchbook that works fast.
The rule: nothing from this stage is presented as a final. It's reference, the same way a mood board of photos is reference.
Step 3: Pick a direction like a human
Out of everything generated and sketched, most gets thrown away. That's the point.
This is the step people skip when they hand everything to a tool. Picking the right direction takes taste, context and a feel for what will still look good in a year. It also takes knowing what the competition already looks like, so you don't end up as a remix of them.
Step 4: Build the real thing by hand
Once the direction is locked, the actual work gets made with the tools I've used for years: Illustrator, Photoshop, Procreate, Figma, Blender, After Effects.
- Logos and lettering get drawn and built as clean vector artwork. Custom lettering is something I care about, and AI is still weak at type. My graffiti typefaces, like Throwie and Fat Finger, were sketched in Procreate and built in FontForge.
- Layouts get designed with real hierarchy, real copy and real brand rules.
- 3D and motion get modeled, lit and animated with intent. On my 3D, motion and AI projects, AI can help with textures or concept frames, but the scene is directed.
Step 5: AI for production speed
Once the creative is set, AI is great at the grind:
- Extending a background to fit a new size
- Removing objects or cleaning up a product shot
- Upscaling an image for a large print
- Making quick variations for testing ad versions
Photoshop's Generative Fill, for example, lets you add and remove content from a photo with a text prompt. That used to be a long, careful clone-stamp job.
Every one of those outputs still gets checked by eye. AI cleanup can leave weird edges, warped text on packaging or extra fingers in the background of a lifestyle shot. Somebody has to catch it.
Step 6: Fit it into the system
A single great image isn't a brand. Each piece has to line up with the guidelines: colors, type, photo style, tone. This is where a lot of AI-heavy work falls apart, because every image was generated on its own with its own look.
When I rewrote the social and brand guidelines at Kicks USA, the point was consistency across a 200+ store retailer's channels. That same thinking applies now. AI makes it easy to produce more content. Guidelines make sure it all still looks like you. My brand guidelines checklist covers what that document needs.
Step 7: Final check and ownership
Before anything ships, I check:
- Does it hold up small, big, in one color, in motion?
- Does it look like anything a competitor already uses?
- Is the final made by a person where it matters?
That last point matters legally too. The U.S. Copyright Office's January 2025 report concluded that purely AI-generated material isn't protected by copyright, while human creative contributions can be. If you want to own your brand assets, a human needs to be doing the creative work that counts. I'm a designer, not a lawyer, so talk to an IP attorney for anything specific.
What I never hand to AI
- The strategy and the big idea
- Logos and custom lettering
- The final call on what's on brand
- Anything with a client's name on it that hasn't been checked by eye
FAQ
How do designers use AI?
Mostly for speed: research, mood exploration, rough comps, background extensions, cleanup and quick variations. Good designers keep strategy, final artwork and brand decisions in human hands.
Does using AI make design cheaper?
It can make parts of the process faster. The thinking, taste and final craft still take real time, and that's where the value of a brand comes from.
Is AI-assisted design lower quality?
Not if someone skilled is directing it and checking everything. It becomes lower quality when AI output gets shipped without that direction.
Will my brand look AI-generated?
Not if the work is built around a clear strategy, custom elements like lettering, and strict guidelines. The generic look comes from skipping those steps.
Want work made this way?
I use AI where it makes your project faster and keep a human hand on everything that defines your brand. If that's the balance you're after, let's talk about your project.