CommonThread AI targets connected, trusted data as the foundation for enterprise AI

Tim Gosnell has a simple diagnosis for why enterprise AI keeps underdelivering: the data is a mess. Not missing, not low-quality necessarily, just scattered across a dozen systems that were never designed to talk to each other. Gosnell, CEO of CommonThread AI, laid this out in an interview with theCUBE's John Furrier, part of the AI Luminaries series sponsored by graph database vendor Neo4j. His argument is that companies have spent years pulling data out of one tool and pasting it into another, losing the relationships between records every time.

That's where graph technology comes in, at least in Gosnell's telling. Relational databases dominate most enterprise stacks, and vector databases have become the trendy add-on for AI retrieval. But graph databases, he says, are built specifically to capture how pieces of data relate to one another, which is exactly what gets lost when teams bounce information between disconnected tools. Go-to-market teams are his go-to example: sales, marketing, and customer success each run their own software, and even when integrations exist, nobody ends up with a shared, coherent view of the customer.

Gosnell frames this as bigger than a plumbing problem. He compares the current moment to the early days of e-commerce, when nobody could have predicted which use cases would actually stick. AI, he argues, is at a similar inflection point, and a lot of its value will come from applications nobody's mapped out yet. Solving data silos isn't just about efficiency in his view, it's about removing the friction that's currently blocking those unimagined use cases from even being tried.

His proposed fix is what he calls a larger context window, not in the narrow AI-model sense, but as an organizational concept: pulling segmentation, ideal customer profiles, messaging, and market intelligence into one continuously updated picture rather than treating them as separate spreadsheets. Feed that context window with economic and political signals, keep refreshing it, and companies get an adaptive read on their own business instead of a snapshot that's stale by the time anyone acts on it. It's a pitch that conveniently maps onto what a graph database is good at, which is worth keeping in mind given Neo4j's sponsorship of the interview series, but the underlying complaint about fragmented enterprise data is one plenty of CIOs would recognize without any prompting.