Flowra · Conversation coaching
A conversation.
A chance to grow.
A desktop app for reviewing conversations, practicing replies and getting live advice. Winner of the Most Useful track award at the Socia Global Build Challenge.
Flowra · Conversation coaching
A desktop app for reviewing conversations, practicing replies and getting live advice. Winner of the Most Useful track award at the Socia Global Build Challenge.
01 · Listen
Flowra captures your microphone and your computer’s audio separately, then turns them into a transcript and a timeline.
02 · Reflect
Review the transcript and get an analysis of what to try next. Gemini, through OpenRouter, powers the AI features.
03 · Practice
Practice and live-assistance views turn the conversation into feedback.
04 · Recognition
Flowra won the Most Useful award at the Socia Global Build Challenge. Built with Majd Alddin Sukkary.

Two-person team · Socia Global Build Challenge · March 2026
A place to reflect on conversations and practice what comes next.
01 / Purpose
Help a person review a conversation, practice responses, and receive assistance while speaking.
02 / Approach
I built the desktop app with Electron, React and TypeScript. Gemini, through OpenRouter, handles analysis, practice, transcription and live advice. Supabase stores the history, and the microphone and system audio are captured separately.
03 / Contribution
Built with Majd Alddin Sukkary. I led the development of the Electron and React desktop app, including the audio capture and transcription flow and the Gemini-powered review, practice and live-assistance features.
04 / Demonstration
Flowra has three flows: transcript review, response practice and live assistance. It labels each line as coming from the microphone or from the system audio.
05 / Results
Flowra won the Most Useful track award at the Socia Global Build Challenge in March 2026.


06 / Limitations
Flowra is a prototype. We haven’t yet measured transcription accuracy, latency or how much it helps real users, and that’s what I’d test next.
07 / Lessons learned
Audio capture, useful feedback, and clear interface states all matter when designing an application around a live conversation.
Next project
The Blender scenes and animations are illustrations made for this portfolio, not recorded results. The YouTube videos show the real robots and lab demos, and the AURORA film includes footage from the interactive learning demo.
The T42 scene uses the Yale OpenHand Model T42 CAD from the Yale GRAB Lab’s OpenHand repository, licensed CC BY-NC 3.0, with the assembly, materials, tendon path and motion adapted for illustration.
R. R. Ma, L. U. Odhner, A. M. Dollar, “A Modular, Open-Source 3D Printed Underactuated Hand,” ICRA 2013.
The Yale OpenHand Project is an initiative to advance the design and use of robotic hands designed and built through rapid-prototyping techniques in order to encourage more variation and innovation in mechanical hardware. Please visit the Yale OpenHand site for more details.