Graphwise Talk#3: Under the Hood of Trustworthy AI: Memory, Meaning, Infrastructure
We started this series with a question about orchestration: why do agentic AI projects keep failing even when the technology is there? The answer, across our previous conversations, has been around the missing piece, which is always around shared, structured understanding that lets agents coordinate, lets auditors trace decisions, and lets organizations actually trust what their AI is doing.
At the Talk #3 with Kurt Cagle, ontologist, RDF architect, and one of the sharpest critics of how enterprise AI is built, we get into what the infrastructure of trustworthy AI actually looks like under the hood: how ontologies encode meaning and how memory states and semantic logs turn probabilistic systems into auditable ones.
This is a conversation-first format, no slide decks. We’ll shape the session around the questions you submit when registering, so don’t forget to send yours in.
What’s on the agenda:
- Why AI that retrieves information and AI that understands it are completely different things, and why that gap is where projects go wrong
- What it means to give AI a memory of your business, not just a search index
- How the organizations getting this right build AI that can explain itself and hold up under scrutiny