I speak about building AI that deserves to be trusted: how agentic systems get designed, evaluated and shipped in a regulated bank, and how an organisation adopts AI without hollowing out its own expertise. Everything I talk about, my team and I have built, run or tested first.
What I speak on
Built by design, driven by evals
How safe agentic systems actually get built in a regulated bank: the architecture sets the shape, and the evals drive the final outcome.
Building with squads of AI agents
New development methods where agents do the building, from workshops where people assemble their own squads.
AI adoption without hollowing out expertise
Making AI organisation-wide without it becoming a checkbox exercise: governance, mental models, and deepening thinking rather than outsourcing it.
What brain science teaches us about AI
A decade of asking careful questions of systems that cannot explain themselves, and why that is the right training for evaluating agents.
Recent
TRC Podcast Series · November 2025
A podcast conversation on making AI adoption organisation-wide without it becoming a checkbox exercise: who should use AI and for what, mental models for working with it, governance and guardrails, and how to deepen expertise rather than outsource thinking.
Microsoft Reactor Sydney, with JobSeeker Circle · February 2024
A panel on AI career paths: roles across research, engineering and everyday use, how to enter or grow in the field, and how generative AI is reshaping work.
The practice behind the talks
Inside Commonwealth Bank I run the fortnightly sessions that keep my chapter current on agentic AI, hands-on workshops where engineers build their own squads of coding agents, and internal hackathons on AI agents. Every talk is tested on working practitioners before it reaches a stage.
Earlier
A speaking record going back to 2006: machine learning, conversational AI and data science seminars at Commonwealth Bank (2017–2018), and a decade of research talks at Stanford, RIKEN and Sydney's universities (2006–2015).