Thoughts on technology leadership, strategy, and digital transformation
Lumeotic is a private place to collect and share photos from trips, parties, and everyday moments with the people who were there.
AI tools help senior and junior engineers in different ways. That changes how managers should think about reviews, learning, and progress.
AI coding tools change what engineers can do. The productivity gains are real, but they move costs rather than remove them.
Board-level AI questions have moved from capability to accountability. Here are four questions technology leaders should be ready to answer.
The main barrier to AI adoption isn't the models or the tooling. It's the data infrastructure underneath. How you sequence that work determines what's actually buildable.
AI agents can move incident response from alerts toward remediation. They help most when the failure is familiar and the signals are good.
AI agents can now respond to some incidents on their own. That changes the on-call role, but it doesn't remove it.
AI tools have made developers faster. That is mostly a good thing. But with more code shipping faster, the operational surface area expands in ways that most organisations are not ready for.
DevOps was built for deterministic code. LLMs are probabilistic and agents act on their own. That changes how we test, observe, and operate them.
DevOps was built for deterministic code. AI agents are probabilistic and autonomous. Operating them requires a different set of checks and guardrails.
When delivery speeds up, compliance gates become a bottleneck or get skipped. Policy as Code helps, but it takes more than installing a tool.
AI tools increase development speed. Traditional compliance gates cannot keep up. The answer is to make non-compliant deployments technically difficult, not merely prohibited by policy.
As AI tools make it easier to write and ship code, the operational surface area grows faster than most teams expect. Platform engineering is how you stay ahead of it — but it's often deprioritised until the cost becomes obvious.
Companies often have the right ideas about what to build. Where things go wrong is the order — addressing the wrong layer first, or solving a future problem before the present one is settled.
The CTO role now sits between engineering, business, and organisation design. The useful skill is often knowing which questions to ask.