How team-zenith-sih-2026/ayusakshi-ip-sakthi-sahayak- works
AyuSakshi is a multilingual, source-cited regulatory guidance platform for Ayurveda IP. Its main workflow takes a practitioner’s jurisdiction-aware chat query through the web client and Node API to the Python RAG service, which retrieves and evaluates legal sources before returning a cited response for storage and display. Separate product paths support formulation classification, ABS/TKDL guidance, document ingestion, authentication, and human facilitator review. The README documents the broader capabilities; unsampled implementation details are represented only where the tree supports a path, and uncertain internal wiring is omitted.
Main entry points: main.py AI API · api.js API client · app.js Express application · rag_orchestrator.py RAG orchestrator
Components 33
User experience
- Web application
client/src/App.jsx - Chat interface
client/src/components/chat/ChatArea.jsx - Classification wizard
client/src/components/classification/ClassificationWizard.jsx - ABS navigator
client/src/components/abs/ABSNavigator.jsx - TKDL checker
client/src/components/tkdl/TKDLChecker.jsx - API client
client/src/services/api.js
Application API
- Express application
server/src/app.js - Chat controller
server/src/controllers/chatController.js - Classification API
server/src/controllers/classificationController.js - AI service client
server/src/services/aiService.js
AI guidance
- AI API
ai-service/app/main.py - RAG orchestrator
ai-service/app/agents/rag_orchestrator.py - Hybrid retrieval
ai-service/app/rag/hybrid_retriever.py - Citation verification
ai-service/app/evaluators/citation_verifier.py - Formulation classifier
ai-service/app/agents/formulation_classifier.py - Classification rules
ai-service/app/agents/classification_tree.py - ABS guidance
ai-service/app/agents/abs_helper.py - Prior-art pointer
ai-service/app/agents/tkdl_pointer.py - Bhashini translation
ai-service/app/multilingual/bhashini_service.py
Knowledge corpus
- Document API
server/src/controllers/documentController.js - Document ingestion
ai-service/app/core/ingestion.py - PDF extraction
ai-service/app/core/pdf_extractor.py - Legal chunker
ai-service/app/core/chunker.py - Ingestion queue
server/src/queues/ingestionQueue.js - PostgreSQL corpus
server/src/config/db.js - Redis
server/src/config/redis.js
Accounts and review
- Authentication
server/src/controllers/authController.js - Facilitator API
server/src/controllers/facilitatorController.js - Facilitator workspace
client/src/pages/facilitator/FacilitatorWorkspacePage.jsx - Knowledge manager
client/src/components/admin/KnowledgeBaseManager.jsx
Other components
- Practitioner
- Human facilitator
- Bhashini service
Connections 39
- Practitioner → Web application: uses
- Web application → Chat interface: renders
- Chat interface → API client: submits query
- Classification wizard → API client: requests classification
- ABS navigator → API client: requests guidance
- TKDL checker → API client: requests prior art
- API client → Express application: calls API
- Express application → Chat controller: dispatches chat
- Express application → Classification API: dispatches classification
- Express application → Document API: dispatches documents
- Express application → Authentication: dispatches auth
- Express application → Facilitator API: dispatches review
- Chat controller → AI service client: requests RAG
- AI service client → AI API: calls AI API
- AI API → RAG orchestrator: runs query
- RAG orchestrator → Hybrid retrieval: retrieves sources
- RAG orchestrator → Citation verification: checks grounding
- Chat controller → PostgreSQL corpus: reads and writes conversations
- Classification API → AI API: calls classification
- AI API → Formulation classifier: classifies formulation
- AI API → Classification rules: walks decision tree
- RAG orchestrator → ABS guidance: ABS capability
- RAG orchestrator → Prior-art pointer: prior-art capability
- AI API → Bhashini translation: translates text
- Bhashini translation → Bhashini service: calls translation service
- Knowledge manager → API client: manages documents
- API client → Express application: uploads documents
- Document API → Ingestion queue: queues ingestion
- Ingestion queue → Redis: uses queue backend
- Ingestion queue → AI service client: requests ingestion
- AI service client → AI API: calls ingest endpoint
- AI API → PDF extraction: extracts pages
- AI API → Legal chunker: chunks text
- AI API → Document ingestion: persists chunks
- Document ingestion → PostgreSQL corpus: writes versions and chunks
- Facilitator workspace → API client: manages cases
- Facilitator API → PostgreSQL corpus: reads and writes cases
- Human facilitator → Facilitator workspace: reviews escalations
- Authentication → PostgreSQL corpus: reads and writes accounts