Your new workforce won't be all human. Join the
conference to learn how to manage it.
request invite
← All conversations
Hanah Darley, Co-founder & Chief AI Officer at Geordi AI — AI adoption, the speed of trust, and everything in between
Hanah Darley
Co-founder & Chief AI Officer at Geordi AI
43:33
September 16, 2025

A path from government intelligence to cybersecurity leadership at Darktrace to AI strategy, and what it tells IT leaders about adopting agentic AI securely.

About the guest
Hanah Darley
Co-founder & Chief AI Officer at Geordi AI

Hanah-Marie Darley is co-founder and Chief AI Officer at Geordie AI, which secures and governs enterprise AI agents. She previously led threat research and AI strategy at Darktrace, and spent close to a decade supporting US federal intelligence operations. She oversaw Darktrace's first published threat reports.

Linkedin
In this episode
00:00 Intro 00:06 Hanah Mary Darley's background in intelligence & AI 02:27 The value of non-linear career paths in AI leadership 03:26 Defining Agentic AI: Large language models with tools 04:52 Governance challenges for IT leaders 06:25 Why security leaders must be at the decision-making table 08:50 Frameworks for AI adoption: ISO, OWASP, NIST, EU AI Act 11:54 Innovation vs. risk: Why IT leaders don't just say "no" 14:34 Explainability vs. performance in AI models 18:26 Jevon's Paradox & the cliff of disillusionment 21:31 Common pitfalls in enterprise AI adoption 24:16 Shadow AI & the need for safe experimentation 27:44 Be kind to your IT leaders: balancing innovation with responsibility 28:59 The future of ITSM: from process orchestration to agent orchestration 32:43 Biggest mistake: treating AI agents like applications 34:51 Designing AI for real human behavior, not "perfect users" 39:48 Human unpredictability in ITSM & AI systems 41:22 Aligning AI capabilities with business outcomes 43:24 Closing reflections & takeaways

Episode transcript

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.