Skip to content
Mobile2026#Mobile#Flutter#AI#Video

GMeeting — AI-transcribed video calls

A cross-platform video conferencing app with real-time transcription, live English-to-Tamil translation, and AI-generated meeting summaries. Runs on Android, iOS, web, and desktop — powered by Agora for video, Deepgram for live speech-to-text, and Groq-hosted LLMs for translation and post-call recap.

GMeeting — AI-transcribed video calls
Client
Studio
Our role
Full-stack Mobile Engineering
Year
2026
Duration
6 weeks
Stack
Flutter · Dart · Agora RTC Engine · Firebase (Auth, Firestore, Storage, Messaging) · Riverpod · go_router · Deepgram WebSocket · Groq / NVIDIA AI · google_sign_in
01 — Challenge

The problem to solve

Online meetings lose context. Tamil-speaking participants miss nuance when calls run in English, action items vanish after hang-up, and teams spend hours rewatching recordings. We needed a video-call app that captured the conversation live, broke the language barrier in real time, and produced a sharable recap the moment the call ended.

02 — Solution

What we built

A Flutter app that streams Agora audio into a Deepgram WebSocket for live captions, pipes each utterance through a Groq-hosted LLM for English↔Tamil translation, and writes every speaker's transcript to Firestore. A live summary panel updates while the meeting runs, and on End Meeting the full transcript is sent back to the LLM to return a structured recap with decisions and action items. Firebase Auth handles Google sign-in, Cloud Messaging triggers reminders, and the same codebase ships to Android, iOS, web, macOS, Linux, and Windows.

03 — Outcome

The impact

Single Flutter codebase across 6 platforms · Live captions render under 400 ms · In-call English↔Tamil translation · Live summary that updates as people speak · Auto-recap delivered seconds after End Meeting · Built end-to-end in 6 weeks.

Gallery

A closer look.

Liked this project?

Let's build something together.

Start a project