The Designer's Guide to Working With AI Without Losing Your Eye

Brandon WatkinsFounder, Flowlabz

The most important thing we've learned about designing with AI is that it will meet you at your current level and help you stay there. If you have a strong point of view, AI amplifies it. If you don't, AI fills the gap with something competent, averaged, and aesthetically inoffensive.

That second outcome is the quiet risk nobody talks about. Not AI replacing designers, but designers slowly outsourcing their taste to a tool trained on the median of existing work, and then wondering why their portfolio feels like everyone else's.

This is a guide for using AI as a genuine accelerator without losing what makes your work yours.

What AI Is Actually Good At in a Design Workflow

Understanding where AI genuinely helps, rather than where it merely appears to help, is the foundation of using it well.

Generative research and moodboarding

AI is excellent at producing a wide array of reference material quickly. Prompting Midjourney or Adobe Firefly to explore visual directions, using ChatGPT to surface competitive examples, asking Claude to describe the design language of a brand concept. These accelerate the research and direction-finding phase significantly.

The key is using these outputs as springboards, not destinations. A Midjourney image that captures the right atmosphere is a reference. It is not the design.

Copy and content at every fidelity

Placeholder copy is one of the most underrated friction points in the design process. "Lorem ipsum" placeholders produce designs that don't actually test the layout with real content lengths. AI-generated copy that sounds like real copy, actual headlines, real-length body text, plausible product names, produces much more honest design feedback.

Use AI to generate content that matches the actual content structure you're designing for. If the client's headlines will be 5–8 words, ask for 5–8 word headlines. If the body copy will be two short paragraphs, generate two short paragraphs. The design should be tested with real-weight content from the start.

Variations and exploration

One of the slowest parts of the design process is generating enough distinct directions to have a genuine choice between them. AI tools can produce 10 layout variations in the time it would take to produce 2 manually. This changes how you work: instead of committing to a direction early and iterating on it, you can generate a wider field of options and select more deliberately.

Where Human Judgment Is Irreplaceable

Point of view

AI has no opinion about your client. It doesn't know that this brand needs to feel authoritative but not cold, or that this product is targeting people who are technically sophisticated but time-poor, or that the last three campaigns in this category have all used the same visual metaphor and this one needs to be different.

Point of view is the designer's core contribution. It comes from understanding the client's context, the competitive landscape, the target audience, and the cultural moment. No amount of prompting produces genuine point of view. You bring that. AI helps you execute it faster.

Taste under constraint

AI is trained on what has existed. It is very good at producing designs that resemble well-regarded existing work. It is not good at producing designs that are right for a specific constraint set. The brand that needs to work at 8px on a wearable, the layout that needs to communicate trustworthiness in a 400ms first impression, the interaction that needs to feel delightful in a product people use under stress.

These are judgment calls that require understanding of the constraint, not just knowledge of design patterns. They are where senior design experience creates value that can't be prompted into existence.

The edit

AI generates abundantly and indiscriminately. It will produce eight options when you asked for three, add a decorative element you didn't ask for, and use a visual metaphor that's technically apt but tonally wrong for the brand. The edit, knowing what to cut, what to keep, and what to push harder, is a design skill that AI does not possess and cannot assist with.

Protect time in your process for the edit. It's where your judgment becomes visible in the output.

Practical Habits for Staying Sharp

Keep a non-AI design practice

Sketching by hand. Designing constraints-first without AI assistance. Studying work you admire and articulating specifically what makes it work. These practices maintain the perceptual muscle that AI use can quietly atrophy if you let it. Think of it as the equivalent of a musician playing without backing tracks. The practice that keeps the skill precise when the tools are removed.

Have an opinion before you prompt

The most effective AI use in design starts with a specific brief, not a vague one. "Generate some landing page options" produces generic output. "Generate a landing page for a B2B SaaS product that targets ops teams in mid-market logistics companies, the tone should feel like a peer talking to a peer, not a vendor pitching, the hero section should show the product, not an illustration" produces something you can actually evaluate against a standard.

The specificity of the prompt reflects the specificity of your thinking. If you can't write a specific brief, you haven't done enough thinking yet. Do that first.

Credit the tool, own the judgment

There's no need to hide AI tool use. It's becoming standard practice. What matters is being clear about what the tool contributed and what you contributed. The tool generated options. You selected, refined, and directed. That selection and direction is the professional service.

Designers who are transparent about their process and confident about the value of their judgment are in a stronger position than those who either hide AI use or let the tool make the decisions. Own your eye. That's what clients are paying for.

Working on a product built by AI tools, rather than designing alongside them? Read how to make an AI-built app look professional.