A scalable intelligence layer built on knowledge graphs is emerging to link structured, unstructured and connected data, turning generic AI models into reliable production systems. According to Jeevan Pathuri, vice president of software engineering at Microsoft Corp., the real value of intelligence comes from connections and relationships across enterprise data, modeled as nodes and edges, which helps language models answer questions more sophisticatedly than standard SQL queries. In Microsoft’s supply chain work, modeling a bill of materials as a graph allowed the team to build 10 to 15 agents in production in a matter of a few weeks, versus building them individually. The reusable semantic layer is where enterprises should invest, with the enterprise knowledge graph market projected to grow from about $1.9 billion today to nearly $10 billion by 2032. Pathuri noted that defining the semantic layer is critical, and an agent that previously took three or four weeks to develop can now be done in a few days. Startups such as Jedify Inc. are raising capital to give agents business knowledge, while the real intelligence stays inside the company, grounded in data relationships.