Neo4j: Knowledge Graphs Give AI Agents Shared Context

Neo4j field CTO Jesús Barrasa argues that enterprises must give AI agents access to shared business knowledge, not just raw data. Without a common knowledge layer, organizations risk repeating past mistakes by embedding context separately into each agent, leading to inconsistent and conflicting results. Barrasa recommends building knowledge graphs incrementally, starting with one use case and aligning each new agent to existing context. Key metrics should include how quickly new agents can be built and the cost of reconciling diverging agent outputs, known as drift.