About

AI agents are moving from demos into production. They’re writing code, triaging cases, and making decisions that touch real people. Yet most teams treat governance as a compliance checkbox, bolted on once the architecture is already defined.

Drawing on his work building AI inside the UK’s data-protection regulator and on national-scale platforms in Africa, Jeremiah makes the case that the person building the agent is the governance layer, and that responsible AI is an engineering discipline you practice in the design itself.

The session lays out six concrete practices for agent builders: how to design systems where traceability, human oversight, and auditability are load-bearing parts of the architecture. Expect an engaging session, with real-life honest trade-offs, and some agent skills you can take back to your own work to codify your AI Governance!

Takeaways

• Why the builder holds the real governance responsibility for an AI system, and what that means for how you architect agents.
• Six practical, repeatable practices for building agentic systems that stay traceable, auditable, and safe to put in front of real users.
• How to make human oversight and auditability load-bearing parts of your design from the start.
• Codifying governance, where we automate the question: “would this agent survive scrutiny?”