Inspiration
What it does## Inspiration
Manual bug reporting is the silent productivity killer of software development. As a student, I realized that developers spend too much time documenting bugs and not enough time fixing them. I wanted to build a bridge between a simple screen recording and a professional technical report.
What it does
Gemini QA Autopilot leverages the multimodal power of Gemini 3 Flash to transform raw video evidence into actionable developer data. It "watches" screen recordings, identifies the exact moment of failure, and automatically generates:
- A professional, detailed bug report.
- An automated Playwright test script to replicate the issue instantly.
How I built it
The core of the system is a Node.js and Express backend. I used the Google Generative AI SDK to process video frames using the Gemini 3 Flash Preview model. The frontend is built with Tailwind CSS for a clean, professional dashboard that focuses on results.
Challenges I ran into
The biggest challenge was optimizing the visual reasoning. Getting the AI to distinguish between a normal UI transition and an actual application crash required precise prompt engineering and careful handling of video frame sequences.
Accomplishments that I'm proud of
I am proud of creating a tool that can "understand" an iOS crash just by looking at a screen recording. Achieving a seamless flow from video upload to a functional Playwright script was a major milestone for this project.
What I learned
I learned a lot about multimodal AI capabilities and how to handle large file uploads in a Node.js environment while maintaining a responsive user experience.
What's next for Gemini QA Autopilot
I plan to integrate the tool directly with Jira and GitHub Issues, and add support for more automation frameworks like Cypress and Selenium.
