Talks

Code was never the product. Understanding was.

AI has made code cheap and fast to produce. Understanding is still slow, and it still lives only in people. Every talk here is about that gap: what it costs, how to close it, and who answers for the decisions made inside it.

Each talk is available as a keynote or a session, in person or virtual, at a conference or inside your organisation. Get in touch to book one, or to shape one for your audience.


AI Literacy

How engineers and teams stay in command when AI writes more of the code. The talks run in an arc: from the metaphors we brought with us, to the personal moment of owning code you can't explain, to the fieldcraft that keeps understanding alive, to the engineers who come next.

Does AI Deliver Waterfall?

Rethinking the metaphors we brought with us.

AI has brought back the industry's oldest metaphor: the factory. This keynote argues the factory was wrong the first time, and is wrong now. Drawing on Peter Naur, cognitive science and centuries of reactions to new technology, it ends with a shift from factory thinking to habitat thinking, where humans and AI each do what they do best.

The Great Engineering Transition: What Happens When Software Starts Thinking?

Rigour doesn't disappear. It moves.

Assembly programmers doubted compilers. Sysadmins feared the cloud. Each time, the craft moved rather than vanished. This transition is different in kind, because the machine now takes part in the reasoning. If the doing is delegated, what is the engineer for?

The Sovereign Engineer: Staying in Command of Code You Didn't Write

How much understanding can you afford to delegate?

The tests pass, attention slips, and weeks later you are responsible for code you can't explain. AI produces code far faster than understanding can be built. Four questions to triage any decision, and a three-stage progression from using AI, to verifying it, to deliberately shaping the environment where decisions are made.

AI Literacy for Developers: Thinking Clearly While AI Writes the Code

The scarce skill is no longer writing code.

When agents can generate whole systems, the scarce skill is reasoning clearly about systems, intent, constraints and feedback loops. How to work with AI without giving up judgement, keep generated systems understandable, and why designing context may now matter more than writing syntax.

Great AI Engineering Habits

The future won't belong to the teams with the biggest models.

It will belong to the teams with the best habits. A fast-paced tour of the behaviours, reflexes and disciplines that separate teams building safe, reliable, explainable AI systems from those releasing unpredictable ones into production.

The Living Harness: Stop AI-Assisted Code from Drifting in Production

Agents forget. Your harness shouldn't.

Every agent session starts with no memory of your conventions, past decisions or the incidents that shaped them, and critical codebases drift quietly. A trust protocol, a five-question interview to surface tacit conventions, a version-controlled harness enforced in your pipelines, and a 30-day plan. Pairs with a hands-on workshop.

The Trust Ladder: How Agentic Systems Earn the Right to Act

Trust is an engineered property, not a feeling from a good demo.

In most organisations, trust in agents is either a hunch or a policy nobody reads. This session engineers it instead: promotion ladders for constraints, permissions classified by reversibility and blast radius, verification at three timescales, and the quiet decay trust suffers when left untended.

Making Agents Responsible for Human Understanding and Judgement

Agents that keep the humans understanding, not just keep producing.

Intelligent platforms promise to remove friction, cognitive load and, eventually, the need to understand what we ship. That last one is not a feature. A personal story of losing the understanding needed to trust my own decisions, and the way back: open-source sentinel agents that review specs adversarially, surface decisions, enforce thin slices and watch for cognitive exhaustion.

The Vanishing Ladder: Growing Engineering Judgment

How do juniors grow judgement when AI does the work they learned from?

For senior engineers and engineering leaders. When AI takes the lower rungs of the apprenticeship, judgement has to be designed rather than absorbed. How juniors learn to spot fragile designs, misleading tests and unsafe assumptions, and how to tell when someone is ready to supervise increasingly autonomous systems.


Decision Sovereignty

One question for any decision an agent makes on your behalf: if this turns out to be wrong, what does it take to put it right?

Lights On Where the Decisions Are

Not the dark factory. Not the dimmed one either.

Few organisations will build software with no one in the room. Most will dim the lights: agents do the work and a person signs at the end, answerable for decisions they were never positioned to understand. Four classes of decision, each with the human attention it deserves, and an environment where reversible work flows freely and consequential decisions reach a person with the evidence to own them.

LGTM Is Not a Decision: Code Review After the Agents Arrived

Approval is not understanding.

Agents produce more code in an afternoon than a team can read in a week, and review has become the place where accountability goes missing. Don't review faster, and don't hand review to another agent. Review decisions, not files, and carry the same lens upstream to the specification and downstream to production.


Platform Engineering and Developer Experience

How Developer Platforms Fail (And How Yours Won't)

Build the platform teams love, not the franken-platform nobody wants to maintain.

Platform engineering's promises tested against real platforms that thrived and struggled, and the design and organisational choices that make the difference.

Developer Experience: Tuning Feedback Loops

You have plenty of feedback. What you lack is loops.

Pipelines, dashboards, alerts, reviews and rituals are not loops. Developer experience as system design, from the workshop floor and Toyota's gemba to modern platform teams, and why AI-generated insights either help or overwhelm depending on when and how they arrive.


Want one of these at your event, or inside your organisation? Get in touch.