Michael Gray

GenAI is a lot like an organisation with low psychological safety.

It will try its best to please you, it won’t challenge you, and when I can’t please you with the truth it will make something up that’s untrue just to satisfy your demands .

Yes I am on a train with too much time in my hands.

All very boring isn’t it

100% this

We need to get better at working with complexity, not flattening it just to feel in control and “productive”. That’s where real judgment, and leadership live.

Anything less isn’t simplification, it’s avoidance. And in my book, that’s piss poor leadership.

It’s easier to follow a familiar metric than to re-engage with the messy, complex reality it once tried to represent.

The real danger? Once a metric’s in place, it builds inertia. It shapes how we talk, what we value, and what we ignore. Even when it’s no longer useful, we keep following it.

“Data-driven decisions” sound smart, but too often it just means letting incomplete or poorly framed data make the decision for us. We forget data is only ever a piece of the picture, useful, but never the whole story.

But in simplifying it all, we lose the nuance that actually matters. We stop thinking deeply, because the numbers look “fine.”

We’ve got this habit, especially in work and leadership, of turning complex, messy, human stuff into clean, objective metrics. It makes things feel manageable, like we’re in control.

“Vibe coding”…

Phrases like this just make me want to get out of tech. My god.

A bit dramatic, I know, but, I don’t love the direction our industry is headed at the moment. Short-termism in full effect. It’s going to be an interesting few years, that’s for sure.

No easy fix, but it’s a cycle that keeps repeating. Would love to hear from folks who’ve seen this done well.

Maybe the answer isn’t big bang migrations but an evolution over revolution approach, gradually modernising and integrating systems instead of assuming a clean break will ever be realistic.

The result? Companies end up supporting multiple systems indefinitely, missing out on the efficiencies they aimed for.

It’s a pattern I see all the time. A company merges, or a new system is built. “All new customers will be on the new platform!” But what about the existing ones? Migration is always underestimated in complexity, cost, and risk. So it drags, or never happens.

Just spent way too long on the phone with EE renewing broadband. The classic story: new customer deals vs. renewal pricing games. But what really stood out? How long it took them to even find my account. Turns out, I’m on their “legacy” system, it took a good 30 minutes of my time.

Psychological safety is NOT about lack of disagreement.

Psychological safety REQUIRES:

* disagreement and debate
* setting standards for behavior and performance, and enforcing them
* telling people things they don't want to hear
* courage, from the bottom up
* humility, from the top down

Thank you for your AI generated response.

“Indeed, AI remains only as good as the data on which it was trained, and the increasing volume of data on the internet being generated by AI risks making it less, not more, reliable.” - and here lies the fundamental flaw with gen AI.

Business/product uncertainty directly increases the complexity of the systems we produce, both software and organisational.

I have no hard evidence this is true but I’m convinced it is.

I can relate, I still hate being told what to do now, especially if I do no understand they why. 😅

With direct interaction, what if you reframe it from advice/instruct, to being curious and asking questions that make them think more deeply. Would that change your perspective on direct interaction?

Why do you think you lean that way?

We should. The other aspect to this is it’s part of our role to ensure the system they work within is not constraining. I.e. they feel they can. If the conditions are wrong we can encourage them all we want, but they won’t be able to.

Global vs local optimisations. The impossible balance 😅

But shaping the future means thinking beyond that, seeing what’s possible before it’s obvious.

Real change happens over time. It’s about shifting mindsets, evolving ways of thinking, and guiding people on a journey.

And the challenge? Explaining why now doesn’t always land well with people, because their mindset is constrained present, not the possibilities of the future.

Most people are focused on today’s problems, immediate priorities, and what they see in front of them.

It’s about thinking strategically, having a perspective on where we should be going, and shaping how we get there.

That means influencing a future that looks different from today.

The premium’s going to be on folks who can design, maintain, and evolve sustainable systems. Big-picture thinkers who connect the dots. These skills are only going to become more and more valuable.

Code is the easy part, especially now that AI is getting better and better at generating it. But AI still can’t fundamentally reason, no matter what the latest ChatGPT claims. And here’s the thing, writing the code was never the hard part.

The is role seeing the business domain clearly, translating it into technical decisions, having strong people and influencing skills, and approaching everything with a systems mindset.

One of my former colleagues asked how important it is for staff/principal engineers to be tech agnostic. My opinion is we should be and largely already are. The real value is understanding fundamentals, they translate to any technology.