Why Your Company Needs a Context Graph (and How to Build It) — Gil Feig, Merge
Your agent can tell you why one customer was upset. Ask which customers were upset last year, and it falls apart. Gil Feig, co-founder and CTO of Merge, explains why every company needs a context graph, a company brain that brings together third-party data, structured context, memories and skills. Using the question "which customers were upset last week?", he shows why connecting a few MCP servers isn't enough: live API lookups can't search across large datasets, so you need a synced copy for semantic queries, alongside live calls for simple lookups. He walks through the architecture: a router, synced and live context, a summarizer, and skills and memory that encode how your company actually defines things like a happy customer. He covers four tiers of context including derived summaries, a decision flowchart for when to use cache or live data, and why every answer must keep track of where its data came from. In this talk: • Why MCP live lookups can't answer questions across large datasets • Combining synced data, live lookups, a router and a summarizer • Why company skills and memory belong in the context layer • Traceability: where data came from, when, and what invalidates it SPEAKER Gil Feig, Co-founder & CTO, Merge LinkedIn: https://www.linkedin.com/in/gilfeig/ X: https://x.com/gilfeig LINKS Merge: https://www.merge.dev Merge Agent Handler: https://www.merge.dev/merge-agent-handler Merge Gateway: https://www.merge.dev/gateway Merge on X: https://x.com/merge_api CHAPTERS 0:00 Intro 1:02 About Merge 2:17 What is a context graph? 2:27 The company brain 3:17 Example: why was a customer upset? 4:02 Why live lookups break 4:22 MCP alone isn't enough 5:07 Sync a copy of the data 6:17 The router and summarizer 6:47 Why skills and memory belong 7:57 No magic, just physics 8:32 Four tiers of context 9:22 Derived context 10:02 The decision flowchart 11:32 Traceability and provenance 12:47 Putting it all together 13:32 Key takeaways 14:17 Merge's products Recorded at the AI Engineer World's Fair 2026 in San Francisco. Subscribe for more talks from the engineers building with AI. AI Engineer: https://ai.engineer YouTube: https://www.youtube.com/@aiDotEngineer X: https://x.com/aiDotEngineer LinkedIn: https://www.linkedin.com/company/aidotengineer/ #ContextEngineering #EnterpriseAI #AIEngineer
More like this

Java 25 killed the boilerplate? void main() explained #shorts

From 15% to 90% GPU Utilization: Fix the Data Pipeline, Not the Model

Small Models, Big Results: Training a Finance Agent for Under $500 — Charles Dickens, Snorkel AI

Generation Is Cheap, Review Is Expensive: How to Stop Shipping AI Slop — Gabriel Martinez, G2i
Join the discussion
Sign in to join the discussion
Sign in