I started with a clear problem I faced every day and turned that into a simple concept for a tool. I designed the complete flow and screens in Figma so I could see exactly how it should look and feel. Then I set up the Figma MCP Server and used the Roo Code(powered by Gemini) inside VS Code to pull the HTML/CSS and then with the help of the agent, converted it to React. I understood the technology of WebRTC and used it for browser-based file transfer, so that files move directly between devices without uploading to a server. Whenever I got stuck or needed help writing extra logic or fixing bugs, I used Gemini 2.5 Pro to get instant answers. Once everything worked, I hosted my frontend on Netlify and my backend server on Render, so the whole tool was live and ready to use.