Zero-Defect Design AI™ Case Study: Avito
In a large design team, design approval takes more time than the work itself. We solved this problem by implementing a custom AI art director, Zero-Defect Design AI™.
The Challenge of Design Review in Large Teams
Reviewing a single layout at Avito required up to five different specialists: senior designers, colleagues from adjacent verticals, design system custodians, and product managers. Each has their own focus, and knowledge is distributed. The result: drawn-out cycles and dozens of hours spent on synchronization. Or worse, skipping important approval stages and having errors surface in production. A single, objective source of truth was needed.
The Task: Teaching AI to Think Like an Art Director
A human art director evaluates a design based on two key criteria: overall design quality and compliance with the design system—checking visual hierarchy, typography, spacing, and accessibility standards. We decided to automate these tasks by teaching an AI to do the same. The neural network is fed a design layout for review, rules from the design system, and a clear checklist for analysis. The output is a detailed and structured error report.
The Quest for Stability
Initial tests on powerful LLM models yielded unstable results. The main problem was a high level of false positives ("hallucinations"), where the AI confidently invented non-existent issues. Standard market solutions have about 30-40% hallucinations; our internal standard was no more than 10%, even on complex mobile layouts.
The Solution
We reduced the false positive rate to below 10% using a special approach: hierarchical mapping of layout elements and structured JSON files that help the neural network navigate the design system.
The "Hidden" Parameter
We also accounted for Avito's imperfect design system. For instance, the model could "hallucinate" an error by not finding a font size that was deeply hidden within the design system's structure. Structured mapping solved this.
After a series of experiments, we settled on the Gemini 1.5 Pro model, which showed the best stability and accuracy. The model effectively understood the context and provided feedback comparable to that of an experienced specialist.
Quality and Predictability
“I liked this review; it’s very useful and really saves time. I saw problems that I would have missed myself.”
“With AI review, the design team started working significantly faster, and releases were postponed less often. On average, the speed of layout approval increased by about 5 times, and the workload on art directors decreased by about 20-25 hours a week, allowing them to focus on more important strategic tasks.”
The Results
Currently, the AI art director operates as a series of prompts and file attachments in a Gemini chat. This is an interim solution to perfectly fine-tune the prompts and process before integrating it into a Figma plugin.
"We reduced the review cycle from several days to 15 minutes, freed up senior designers' time, and improved the quality of the final product. The entire implementation process took only three meetings."
Scalable Expertise
Art directors' knowledge is available 24/7.
Company Taste
The AI works according to the criteria and perspective of the current art director.
Speed and Focus
80% of feedback in minutes, allowing more time for strategy.
Customized Process
The system is tailored to the company's specific review processes.
This approach is the new standard for large teams.
It saves the most valuable asset — the time of your best specialists. Want the same?