How kathirvbk/ai-meet works
MeetMind’s principal workflow accepts a meeting recording or transcript, processes speech into speaker-aware text, runs a validation-loop agent to produce minutes and action items, and stores meeting records for later access. The FastAPI service also exposes history, administration, and meeting Q&A; the React frontend provides the user-facing meeting and management screens. The README describes a Streamlit UI, but the supplied tree instead contains a React frontend and FastAPI server, so the graph follows the files present. Some subsystem links rely on the README or SOURCE INDEX; unsampled implementation details are not inferred.
Main entry points: api_server.py API server · transcript_service.py Transcript pipeline · workflow.py Agent workflow · validator.py Output validation
Components 26
Web client
- React app
frontend/src/App.jsx - Meeting screens
frontend/src/pages - Admin screens
frontend/src/pages/AdminPanel.jsx - API client
frontend/src/utils/api.js
API and services
- API server
api_server.py - Meeting history
services/history_service.py - Admin services
services/admin_service.py - Relational models
database/models.py - PDF export
services/pdf_service.py
Speech processing
- Audio processing
services/audio_processor.py - Speech recognition
services/speech_to_text.py - Speaker diarization
services/speaker_diarization.py - Speaker naming
services/speaker_name_detector.py - Transcript reconstruction
services/transcript_reconstructor.py - Transcript pipeline
services/transcript_service.py
Minutes workflow
- Agent workflow
graph/workflow.py - Meeting analysis
agents/meeting_analyzer.py - Action extraction
agents/action_extractor.py - Output validation
agents/validator.py - Error correction
agents/reanalyzer.py - Minutes generation
agents/mom_generator.py
Meeting knowledge
- Club Q&A
rag/rag_chain.py - Meeting retrieval
rag/retriever.py - Vector store
rag/vector_store.py
Other components
- Club user
- Groq AI services
Connections 34
- Club user → React app: uses
README.md - React app → Meeting screens: routes to
frontend/src/App.jsx - React app → Admin screens: routes to
frontend/src/App.jsx - Meeting screens → API client: calls
frontend/src/pages/MeetingDetails.jsx - Admin screens → API client: calls
frontend/src/pages/AdminPanel.jsx - API client → API server: HTTP requests
- API server → Transcript pipeline: invokes
api_server.py - API server → Agent workflow: invokes
api_server.py - API server → Meeting history: invokes
api_server.py - API server → Club Q&A: queries
api_server.py - API server → Admin services: invokes
api_server.py - Meeting history → Relational models: persists records
services/history_service.py - Admin services → Relational models: uses
services/admin_service.py - Transcript pipeline → Audio processing: converts and chunks
services/transcript_service.py - Transcript pipeline → Speech recognition: transcribes
services/transcript_service.py - Transcript pipeline → Speaker diarization: identifies speakers
services/transcript_service.py - Transcript pipeline → Speaker naming: detects names
services/transcript_service.py - Transcript pipeline → Transcript reconstruction: reconstructs text
services/transcript_service.py - Agent workflow → Meeting analysis: dispatches
graph/workflow.py - Agent workflow → Action extraction: dispatches
graph/workflow.py - Agent workflow → Output validation: dispatches
graph/workflow.py - Agent workflow → Error correction: dispatches
graph/workflow.py - Agent workflow → Minutes generation: dispatches
graph/workflow.py - Meeting analysis → Action extraction: then extracts
README.md - Action extraction → Output validation: then validates
README.md - Output validation → Error correction: retries errors
README.md - Error correction → Output validation: revalidates
graph/workflow.py - Output validation → Minutes generation: continues when valid
README.md - Minutes generation → Groq AI services: generates minutes
- Club Q&A → Meeting retrieval: retrieves context
rag/rag_chain.py - Meeting retrieval → Vector store: searches
rag/retriever.py - Vector store → Groq AI services: embeds records
- Club Q&A → Groq AI services: generates answers
rag/rag_chain.py - Club user → PDF export: downloads minutes