Back to case studies
Requirements Copilot and Knowledge Base with Temporal Graph RAG
Client name withheld · Consulting
Requirements
RFPs and transcripts
traceable hierarchies
Retrieval
vector + temporal graph
Access
MCP · OAuth · rate limits
A professional-services firm's RFPs and meeting transcripts, turned into traceable requirement hierarchies by a copilot built on temporal graph RAG.
What changed
Before
Sources
RFPs
Meeting transcripts
After
Retrieve
Weaviate
vector · HyDE
Graphiti + Neo4j
temporal graph
Model
Claude on Bedrock
Agent SDK · Titan v2
Serve
FastAPI knowledge service
Access
MCP server
OAuth · rate limits
RFPs and transcripts go in; vector and temporal graph retrieval turn them into requirement hierarchies, and an authenticated MCP server makes them queryable.
What we built
The copilot turns RFPs and meeting transcripts into requirement hierarchies with temporal graph RAG. A FastAPI knowledge service runs vector search with HyDE query rewriting and serves an authenticated MCP server with OAuth and rate limiting.
Key Features
Temporal graph RAG: Graphiti on Neo4j, beside vector search in Weaviate.
A FastAPI knowledge service that rewrites queries with HyDE before the vector search.
An authenticated MCP server, with OAuth and rate limiting.
Claude on Amazon Bedrock and the Claude Agent SDK, with Titan v2 embeddings.
SQS FIFO queues and a Streamlit app.
Results
RFPs and meeting transcripts become requirement hierarchies that trace back to their source.
AI assistants reach the knowledge base through the authenticated MCP server.
Stack
Engagement: 6 months · AI-Assisted Engineering, AI Infrastructure & MLOps
Next step
Let’s look at your SDLC together.
45 minutes with a senior engineer: your setup, the risks, what to fix first. No sales team.