McKinsey's James Kaplan: AI turns messy data into knowledge graphs

Kaplan framed knowledge graphs as familiar technology that enterprises have been slow to adopt. He noted that social media companies arrived at the power of graphs before the enterprise did, and that anyone using LinkedIn, Wikipedia or social media is already using a graph. The shift he described is about who does the work: previously, evaluating messy information required time-consuming, expensive and often imperfect work by business analysts or data scientists. Now, he said, AI can interrogate that data and create deterministic business rules programmatically.

McKinsey's own use of the technology runs through EcliptOS, which the firm describes as an AI operating system that connects C-suite strategy with everyday execution through agentic workflows. Kaplan said the system amounts to a graph of databases, and that graphs' flexible data schemas make it easier to create a virtual graph connecting many databases. He also suggested business priorities should guide where organizations apply AI improvements, giving customer experience as a possible example ahead of productivity. The richer the interconnections among nodes, he said, the more intelligence the graph has and the more things can be determined.