Why Your Company Needs a Context Graph (and How to Build It) — Gil Feig, Merge

Why watch

Gil Feig explains why AI agents need a context graph architecture to answer complex business questions across large datasets. He demonstrates why live MCP API lookups fail for queries spanning historical data, and how combining synced semantic data with live lookups, routers, and summarizers creates a powerful company brain. The talk covers four tiers of context, decision frameworks for cache vs live data, and the critical importance of tracking data provenance for trustworthy AI responses. Ideal for AI engineers and developers building agent systems who need to handle complex, multi-source queries.