AI Design Reviewer – Project Story

Inspiration

As a developer, I noticed that good design feedback is often expensive, slow, or unavailable to early-stage founders and students. Many teams launch products with usability, accessibility, or cultural design issues simply because they lack access to experienced UX designers. I was inspired to explore whether a large reasoning model like Gemini 3 could simulate the thinking process of multiple design experts and provide meaningful, actionable feedback instantly.

This led to the idea of building an AI-powered design reviewer that could analyze a website or design description and deliver professional-quality insights.

What I Learned

Through this project, I learned how to use Gemini 3 beyond basic text generation. I explored expert role prompting, structured output generation, and multi-step reasoning within a single request. I also learned how careful prompt design can guide the model to assign severity levels, prioritize recommendations, and avoid generic advice.

This project deepened my understanding of how large language models can act as decision-support systems rather than simple chat interfaces.

How I Built the Project

I built the application using Gemini AI Studio, focusing on a fast and reliable prototype. The app accepts a website URL or design description along with the target audience and country. Gemini 3 is prompted to act as a senior UX designer, accessibility specialist, and product consultant.

The model performs a structured analysis across usability, accessibility (aligned with WCAG principles), cultural design fit, and content clarity. The output is organized into clear sections with prioritized improvement suggestions, making the feedback immediately actionable.

Challenges Faced

One major challenge was preventing overly generic responses. I addressed this by refining the prompt to enforce structure, severity levels, and explicit prioritization. Another challenge was balancing depth with clarity so that the output remained professional but easy to understand within a short review.

Overall, this project demonstrated how Gemini 3’s reasoning and contextual understanding can be leveraged to deliver expert-level insights at scale.

Built With

• accessibility
• and-cultural-design-evaluation-web-based-ui-(ai-studio-apps)-?-lightweight
• and-deploy-the-interactive-application-quickly-prompt-engineering-?-custom-system-prompts-to-guide-gemini-3-in-ux
• and-structured-feedback-generation-gemini-ai-studio-?-platform-used-to-build
• design-analysis
• gemini
• geminiaistudio
• promptengineering
• test