Sharing all KGC 2026 decks. More production-grade KG systems than I’ve seen at any conference. [D]

Didn’t make it to New York for the Knowledge Graph Conference this year, but caught some talks virtually and managed to download all the decks. Sharing them below because some of what was shown is worth knowing about.

Majority of the presentations described live production systems. Enterprises showing up with real engineers delivering real compliance requirements. That’s not usual for most ai eventss. Most talks are proofs of concept with a "coming soon to prod" slide at the end.

For eg – Bloomberg showed a formal dependency model for ontology governance. AbbVie walked through ARCH, their internal KG for drug and disease-area intelligence, connected to a scoring engine, a researcher dashboard, and an LLM companion for plain-language queries. The KG is the source of truth. The LLM is the interface. Even Morgan Stanley showed continuous SHACL drift detection on risk reporting data – automated weekly checks that alert when the semantic layer deviates from what’s governed.

Crux: knowledge graphs are being actively used as infrastructure, not a retrieval layer on top of vectors. The graph is doing reasoning work, not lookup work.

We’ve been skeptical of the "only using vector dbs" framing for a while. These production systems are the clearest evidence I’ve seen of where that breaks down – and what the alternative actually looks like when it’s running. Link to the all the decks in the comment.

All decks here:

https://drive.google.com/drive/folders/1Csdv4hZePrBMJGggsisPXYBueTRCK1kV?usp=sharing