Skip to main content

Thinking

I wanted a way to bring together some of the more specific things that I am currently thinking about which might otherwise get lost in the rolling content of the blog.

These are the things that I keep coming back to — the problems that don't have easy answers but probably have better ones than many teams are currently finding.

The main things firing my synapses right now are:

Speed and quality are not opposites​

Speed and quality are not a dial you turn up and down — that's a false trade-off that leads to bad decisions.

The real question is whether your team has the engineering judgement, the tooling, and the psychological safety to make good calls quickly — and to learn from bad calls.

Most teams that think they have a speed problem actually have a trust problem, or a clarity problem, or both.

Technology problems are almost never technology problems — they're almost always people, process, or time problems.

AI changes leadership more than it changes tooling​

Most of the conversation about AI in engineering is about tooling.

Almost none of it is about what changes for the people leading those teams — how you maintain engineering judgement when juniors can ship faster than seniors can review, how you govern quality without becoming a bottleneck, and what "good" even looks like when the fundamental unit of work has changed.

That's the bit I find most interesting.

Empowerment requires accountability, and accountability requires support​

Most engineering teams don't fail because the engineers are bad — they fail because the business imperatives are wrong, or at odds with the logical process of technical implementation. They lack empowerment, or they have been empowered without accountability and — most importantly — support.

Engineers (and I may be projecting here...) need time to understand a problem space, design a solution, implement it, figure out where it was wrong, fix it, and move forward.

Business people want neat, predictable delivery every time despite a vast field of unknowns existing between here and there.

Finding the right balance of psychological safety and the ability to commit to meaningful delivery is the secret sauce.

Empowerment without support is just abandonment with better branding.

Judgement is becoming the scarce resource​

AI is rapidly commoditising knowledge and execution.

That doesn’t make engineers less valuable — it changes where the value sits.

The differentiator is increasingly:

  • Framing the problem
  • Making trade-offs
  • Knowing when the output is wrong

Most organisations are still optimising for throughput when they should be optimising for judgement.

We are optimising away our own capability pipeline​

AI allows smaller, more senior-heavy teams to deliver more.

That’s efficient in the short term.

But it quietly removes the mechanisms that create future senior engineers:

  • Fewer junior roles
  • Less exposure to real complexity
  • Narrower progression paths

You don’t feel that immediately.

You feel it later, when you realise you don’t have enough people who can make the hard decisions.

Time is no longer consistent across the system​

Different parts of an organisation now operate at different speeds:

  • AI and tooling change weekly
  • Delivery operates in sprints
  • Capability develops over years
  • Strategy is still often annual

That misalignment creates friction everywhere:

  • Plans drift
  • Skills lag
  • Output increases without alignment

The problem is no longer just speed.

It’s how you design systems that can operate across multiple, conflicting time horizons.

Last updated on