Context Graph Ingest
Overview
A harness for turning documents, talks, repositories, profiles, and transcripts into a source-labeled graph that agents can retrieve without losing provenance.
Conference Context
- Context graph, GraphRAG, memory, and retrieval sessions emphasize relationships rather than isolated chunks.
- The wiki already separates official schedule facts, supporting videos, transcripts, OCR, and public source-of-source context.
- Search and memory talks show that source reachability and relationship quality are part of agent performance.
Implementation Pattern
- Ingest sources by source type and preserve the original file or URL for every claim.
- Extract entities, relationships, and claim snippets into separate reviewable records.
- Score retrieval by whether it returns the right source for the question, not only similar text.
- Keep stale or weak sources reachable but labeled, so agents do not treat every edge equally.
- Promote stable graph patterns into topic pages, evaluations, or playbooks only after review.
Evidence
- agent memory - Topic synthesis
- agentic search - Topic synthesis
- what context graph and memory architecture is practical - Question layer
- 2026 06 30 gil feig why your company needs a context graph and how to build it - Official schedule
- 2026 06 29 nyah macklin rag needs a map using graphrag to retrieve connected context - Official schedule
- 2026 07 01 daniel chalef citation needed provenance for llm built knowledge graphs - Official schedule
Evidence Boundary
This is a reusable workflow synthesized from the linked conference evidence. Treat it as a recommended implementation pattern, not as a direct quote from any single talk.