How jeev1992/supportdesk-rag-workshop works
SupportDesk-RAG is a hands-on workshop, not one integrated runtime application: its six independently runnable demos teach a progression from embeddings and document preparation through retrieval, answer generation, evaluation, and tool-using agents. The graph preserves those distinct learning stages and the agent’s demonstrated tool-dispatch layer. README statements establish the workshop capabilities; source excerpts confirm selected implementation details. Other module internals and cross-module data flow were not inspected, so no runtime pipeline between demos is implied.
Main entry points: 4_rag_pipeline/demo.py RAG pipeline · 5_evaluation/demo.py RAG evaluation · 1_embeddings/demo.py Embedding similarity · 2_chunking/demo.py Document chunking
Components 14
Retrieval Foundations
- Embedding similarity
modules/1_embeddings/demo.py - Document chunking
modules/2_chunking/demo.py - Index strategies
modules/3_indexing/demo.py
RAG Applications
- RAG pipeline
modules/4_rag_pipeline/demo.py - Ticket retrieval
modules/4_rag_pipeline/demo.py - Grounded generation
modules/4_rag_pipeline/demo.py
Evaluation and Agents
- RAG evaluation
modules/5_evaluation/demo.py - Retrieval metrics
modules/5_evaluation/demo.py - Generation metrics
modules/5_evaluation/demo.py - Agentic RAG
modules/6_agentic_rag/demo.py - Tool dispatch
modules/6_agentic_rag/demo.py - Ticket tools
modules/6_agentic_rag/tools.py
Other components
- Workshop learner
- OpenAI APIs
Connections 17
- Workshop learner → Embedding similarity: runs demo
README.md - Workshop learner → Document chunking: runs demo
README.md - Workshop learner → Index strategies: runs demo
README.md - Workshop learner → RAG pipeline: runs demo
README.md - Workshop learner → RAG evaluation: runs demo
README.md - Workshop learner → Agentic RAG: runs demo
README.md - Embedding similarity → OpenAI APIs: requests embeddings
modules/1_embeddings/demo.py - Document chunking → OpenAI APIs: embeds sentences
modules/2_chunking/demo.py - RAG pipeline → Ticket retrieval: configures
modules/4_rag_pipeline/demo.py - RAG pipeline → Grounded generation: generates answers
modules/4_rag_pipeline/demo.py - RAG pipeline → OpenAI APIs: uses models
modules/4_rag_pipeline/demo.py - RAG evaluation → Retrieval metrics: evaluates retrieval
modules/5_evaluation/demo.py - RAG evaluation → Generation metrics: evaluates answers
modules/5_evaluation/demo.py - RAG evaluation → OpenAI APIs: requests judgments
modules/5_evaluation/demo.py - Agentic RAG → Tool dispatch: dispatches calls
modules/6_agentic_rag/demo.py - Tool dispatch → Ticket tools: invokes selected tool
modules/6_agentic_rag/demo.py - Ticket tools → OpenAI APIs: creates embeddings
modules/6_agentic_rag/tools.py