How navzar81-dev/talk---a-decision-making-model-leveraging-laya-and-jev- works
TALK is a voice-first decision assistant: a user speaks or enters a dilemma, the backend triages and frames it, Laya scores the choices, and the interface presents calibrated reasoning and can speak or export the result. Research-dependent decisions add LLM framing and web evidence; document grounding, configurable rules, provider settings, and voice personas are supporting capabilities. The README and source index establish the main calls and capabilities; some frontend-to-API wiring and unsampled subsystem internals are not shown, so those links are omitted.
Main entry points: App.tsx Workspace controller · main.py Backend API · orchestrator.py Decision orchestrator · llm_manager.py LLM framing
Components 21
Decision workspace
- Workspace controller
frontend/src/App.tsx - Liquid orb
frontend/src/components/LiquidOrb.tsx - Voice orb
frontend/src/components/VoiceOrb.tsx - Orb visualization
frontend/src/components/Orb.tsx - Audio analyzer
frontend/src/utils/audio.ts - Decision results
frontend/src/components/RightPane.tsx - Versus builder
frontend/src/components/VersusModal.tsx - Provider settings
frontend/src/components/SettingsModal.tsx
Decision pipeline
- Backend API
backend/app/main.py - Decision orchestrator
backend/app/orchestrator.py - Rule classification
backend/app/classification_service.py - Laya scoring
backend/app/laya_service.py
External services
- LLM framing
backend/app/llm_manager.py - Voice synthesis
backend/app/voice_service.py - LLM providers
- Web knowledge index
- Neural voice services
Grounding and output
- Badge exporter
frontend/src/utils/cardExporter.ts - Document parser
backend/app/document_parser.py - API data models
backend/app/models.py
Other components
- Decision maker
Connections 20
- Decision maker → Workspace controller: enters decision
README.md - Workspace controller → Liquid orb: renders
frontend/src/App.tsx - Liquid orb → Voice orb: renders
frontend/src/components/LiquidOrb.tsx - Workspace controller → Audio analyzer: uses audio
frontend/src/App.tsx - Workspace controller → Decision results: renders results
frontend/src/App.tsx - Workspace controller → Versus builder: opens duel
frontend/src/App.tsx - Workspace controller → Provider settings: opens settings
frontend/src/App.tsx - Decision results → Badge exporter: exports badge
frontend/src/components/RightPane.tsx - Workspace controller → Backend API: submits decisions
- Backend API → Decision orchestrator: routes decision
backend/app/main.py - Backend API → Document parser: parses upload
backend/app/main.py - Backend API → Voice synthesis: requests speech
backend/app/main.py - Decision orchestrator → Rule classification: classifies query
backend/app/orchestrator.py - Decision orchestrator → LLM framing: frames research
backend/app/orchestrator.py - Decision orchestrator → Laya scoring: scores choices
backend/app/orchestrator.py - LLM framing → LLM providers: queries provider
README.md - LLM framing → Web knowledge index: searches evidence
backend/app/llm_manager.py - Voice synthesis → Neural voice services: streams speech
README.md - Backend API → API data models: validates payloads
backend/app/main.py - Decision orchestrator → API data models: builds responses
backend/app/orchestrator.py