Knowledge graph architecture gives enterprises ownership of the AI intelligence they create
Knowledge graph architecture has emerged from the concept stage, consigned to the realm of academia, to the production infrastructure stage, and the enterprises that shape it the right way will own the intelligence their AI creates.
That’s the central premise Shan Rizvi (pictured), founder and context architect at Thumos Care, is offering to practitioners who still treat AI retrieval as a search issue. Retrieval-augmented generation finds chunks of text, but it doesn’t reason, trace decisions back to evidence or accumulate knowledge over time. The rift between organizations that understand this and those that don’t will become one of the defining competitive divides of the agentic era, Rizvi noted.
“When you make an argument, there’s a premise, some reasoning applied to the premise, and then a conclusion,” Rizvi said. “Structure is quite inherent to thinking, and graphs, with the right ontology, are the obvious substrate.”
Rizvi spoke with theCUBE’s John Furrier for an exclusive AI Luminaries interview series on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed why knowledge graph architecture is becoming the essential intelligence layer for enterprise AI, how to architect it across structured and unstructured data sources and what success actually looks like for teams trying to build a durable company brain. ( Disclosure below.)*
Knowledge graph architecture enables traceable reasoning that enterprises own and control
The practical case for knowledge graph over search boils down to traversal. Rizvi described a startup building AI for truck parts maintenance. A system that needed to find all substitutes for a given part manufactured by companies meeting multiple criteria simultaneously. Search struggles with this kind of relational path-following at scale throughout millions of SKUs. A knowledge graph deals with it natively and can repeat the traversal recursively throughout different paths until the best answer materializes. This mirrors how human reasoning occurs, rather than how information retrieval has historically been engineered.
“That type of traversal is quite critical to reasoning,” Rizvi said. “Recursion is another key component. You might need to traverse certain paths in the graph, then repeat that traversal in different directions to find the best piece of information.”
For production deployment, the number one consideration is latency, Rizvi noted. Building a medical knowledge graph for Thumos Care using Neo4j — connecting diseases, symptoms, drugs and clinical guidance at scale — revealed that not all graph database platforms handle millions of nodes reliably. Recursive reasoning should only engage when necessary, with faster single-hop retrieval handling simpler queries to keep response times viable in production. “Scalability is quite critical to consider,” Rizvi said. “If you’re dealing with millions of parts and tens of millions of SKUs, you need a graph database platform that can handle that reliably at a reasonable latency.”
On the question of data infrastructure, Rizvi drew a clear line between what belongs in a graph and what doesn’t. Structured, traditional databases with frequently changing data should remain in their native systems, with only their ontological structure represented in the graph so agents know what to query and when. Unstructured sources like Slack threads, meeting transcripts and chat logs are better candidates for direct extraction and ingestion into the graph, where their contextual meaning can be preserved and linked to domain entities over time.
“If you want to own a piece of the intelligence ecosystem, the piece you own has to be seen as intelligence, not as data,” Rizvi said. “You need an ontology for your domain, for knowledge, for cognition and for decision-making, and as long as you have those components, your models will get better as they’re used more frequently, and you will own that value.”
Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s exclusive coverage of the AI Luminaries interview series:
( Disclosure: TheCUBE is a paid media partner for the Neo4j AI Luminaries interview series. Neither Neo4j, the sponsor of theCUBE’s coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)*
Photo: SiliconANGLE
Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities.
15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more** 11.4k+ theCUBE alumni**— Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network.
About SiliconANGLE Media
theCUBE AIand theCUBE SuperStudios — with flagship locations in Silicon Valley and the New York Stock Exchange — SiliconANGLE Media operates at the intersection of media, technology and AI.
Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.