How to Use Gemini for Product Design and UX Work
Gemini is Google's family of AI models, available through the Gemini app and Google AI Studio.
The reason it earns a place in a design workflow is that it reads images properly. You can hand it a screen and ask it to reason about what it is looking at, which is closer to how design review actually works than typing a description of your own work.
Analysing a screen
Upload a screenshot and ask what the eye sees first, second and third, and where attention gets lost.
Then the useful follow up: is that the order I want. This takes a minute and catches the hierarchy problems you have gone blind to after a week in the file.
Catching accessibility problems early
Ask it to review the screen for contrast, text size, tap target spacing and unclear labels.
It will not replace a real audit and I would not present its output as one. It catches the obvious problems while they are still cheap to fix, which is most of the value of an audit anyway.
Microcopy
Headlines, button labels, onboarding, empty states. Give it the user's situation and ask for several options at different tones, then push the best one into shape.
The second round is where the quality comes from. The first round is raw material.
Research and competitive work
Summarising competitor flows, clustering feedback into themes, drafting research questions. Its position inside Google's ecosystem makes gathering and organising this quick.
Early exploration
Good for breadth. Many rough directions fast, none of them finished.
Use it to get off the blank canvas, then stop using it once you know which direction you are working in.
What makes the difference
- Show it images rather than describing them. That is the thing it is best at, and describing your own design to it wastes the advantage.
- Give context and intent: who the user is, what the business needs.
- Ask for options and tradeoffs rather than one answer.
- Treat everything as a first draft and put your own taste on top.
Questions I get asked
Can it really analyse screenshots? Yes, and that is its most useful property for designers. Upload a mockup and ask it to reason about the visual design.
Can it replace a designer? No. It speeds up production and analysis. Deciding what to build, and whether it serves both the user and the business, is still yours.
Gemini or something else? Gemini is strongest on multimodal, image based tasks. Most designers I know use more than one tool and pick per stage rather than picking a favourite.