About

Understanding what a trained model actually is, what it has learned, and what it is capable of is essential to using AI effectively. Rather than starting with a list of AI use cases, this session starts with the model itself and builds an intuition for where these systems are genuinely useful within a data estate.

Most organisations aren’t running on a greenfield data stack. They’re running on legacy on-prem systems, years of accumulated business logic and pipelines that quietly keep the organisation moving. The challenge is working out where AI can add value without assuming you can replace the systems already in place.

Drawing on experience embedding AI into an established enterprise data platform, I’ll explore how a practical understanding of trained models can help identify opportunities across a mature data estate, including automated documentation and lineage, knowledge governance, drift detection and surfacing business logic hidden in stored procedures and tribal knowledge.

Key Takeaways:

• How to identify practical AI opportunities within an existing data estate
• How LLMs can help uncover and document business logic trapped in legacy systems
• How to introduce AI without giving models access to live or sensitive data
• The governance considerations that matter when applying AI in regulated environments