Jon Cooke

AI agents transform how organisations operate, but in regulated industries the question is never just "can it work?" - it is "can you prove it, audit it, and stand behind it?" Most agent frameworks are built on probabilistic foundations that make this a hard problem space. This session introduces designing determinism into agents from the ground up, so that every decision is explainable, every action is bounded by domain constraints, and every outcome traces back to a source regulation, guideline, or policy. Jon Cooke, Founder of Nebulyx.AI, will show how Cognition, a consistent AI engine, combined with AgentDOG, an adaptive agent framework built on a Data Object Graph, makes this possible - purpose built for the compliance, sovereignty, and auditability demands of regulated industries. Key Takeaways: - Learn why separating knowledge grounding from agent execution eliminates hallucination risk while giving agents bounded decisioning that learns and adapts through execution. - Discover how to encode regulatory domain knowledge with 70-80% captured automatically from source, replacing months of upfront work whilst keeping SMEs in control. - Understand how Data Object Graphs deliver fully auditable agent decisions with complete provenance and deterministic execution

How Generative AI dynamically transforms business problems into executable graphs of data products, by creating AI-powered digital twins. Describing business needs in plain language and generating and simulate entire end to end business processes with UX as an interconnected eco-system as graph data products, demonstrated through real-world use-cases.