McKinsey Links Enterprise Data Through Knowledge Graphs to Give AI Business Context

The conversation was part of theCUBE + NYSE Wired: AI Luminaries interview series, broadcast exclusively on theCUBE, SiliconANGLE Media's livestreaming studio.

Kaplan said the underlying technology is already familiar to most people, even if they do not recognize it as a graph. "If you're using LinkedIn, if you're using Wikipedia, if you're using any social media, that's a graph," he said, adding that many social media companies reached the technology before enterprises did. He described graphs as more intuitive than relational databases, which he said work well for transactional data but are much less suited to ambiguous or complicated information. Describing a customer, a product, a process or a step in that process in relation to other things, he said, is where the approach becomes insightful.

According to Kaplan, AI can help convert unstructured information into structured data and business rules that can be stored in a knowledge graph. That work previously fell to business analysts or data scientists and was time-consuming, expensive and often imperfect. "Now, for the first time, we have the ability to interrogate complicated processes and create deterministic business rules programmatically," he said, describing the ability to turn messy, uncorrelated, unstructured data into structured data stored in a graph as opening new frontiers.

Kaplan said business priorities should guide where organizations apply AI improvements, and that customer experience may take precedence over productivity. "What if I pointed these AI improvements at customer experience rather than productivity?" he asked. "The richer the interconnections among nodes, the more intelligence you have in the graph and the more things you can determine."

McKinsey uses AI and knowledge graphs through EcliptOS, an AI operating system the firm says is designed to connect C-suite strategy with everyday execution through agentic workflows. The system relies on a semantic data layer that organizes data and its relationships to support generative AI applications. "What we in effect created was a graph of databases," Kaplan said. "One of the nice things about graphs is they have more flexible data schemas. It's easier to create a virtual graph that connects many databases. And that to me is incredibly powerful."

TheCUBE is a paid media partner for theCUBE + NYSE Wired: AI Luminaries interview series. Neo4j, the sponsor of theCUBE's event coverage, and other sponsors have no editorial control over content on theCUBE or SiliconANGLE, according to the report.

Editor's Summary

McKinsey partner James Kaplan said knowledge graphs give enterprise AI the relationships between customers, products and processes that relational databases handle poorly, and that AI can now convert unstructured information into structured data and deterministic business rules. McKinsey applies the approach through EcliptOS, an AI operating system built on a semantic data layer, with Kaplan arguing that customer experience rather than productivity may deserve the first wave of such improvements.