Search is Back: Solving the "Context Crisis" for AI Agents
summary
Shows how to combine Knowledge Graphs + Vector Search with open source tooling (Cogni, Quadrant, TurboPuffer) to give agents structured context, not just chunks. Proposes session vs permanent memory where session traces are summarized into permanent nodes, storing decision traces (prompts, tools, outcomes) so agents learn precedents. Demoed a remember() quickstart that builds a graph in minutes, claims the default pipeline fits 90% of use cases, and reports pilots across 100+ companies.
AI-written summary · about
chapters (4)
- 0:00Introduction and context problem
- 10:15Evolution of search and context
- 20:48Cogni’s memory platform approach
- 30:12Q&A on contradictory data and knowledge graphs
related talks
26mStore Your AI Agent's Memory and Context in a Knowledge Graph
29mThe AI Agent Memory Landscape
29mHow to Structure Context to Make Your Agents Smarter
23mConnecting the Dots With Context Graphs
32mOpenSearch Agentic Memory: Intelligent Context for AI Agents
26mOpening Keynote: Exploring Context Graphs: From Data to Decisions
30mBeyond Keywords: Building an AI-Powered Search That Reads Your Mind
31mBuilding the Context Engine AI Agents Need