AI Won't Replace Your Engineers. But It Will Expose Your Lack of Engineering Leadership.

AI Won’t Replace Your Engineers. But It Will Expose Your Lack of Engineering Leadership.

Every week there’s a new prediction. “AI will replace 80% of software engineers.” “Coding is dead.” “The last developer has already been born.” And every week, the actual data tells a completely different story.

CNN reported this month that software engineering job listings are up 11% year-on-year. The Bureau of Labor Statistics projects 15% employment growth through 2034. Companies like Block are using AI to make each engineer dramatically more productive, but they’re not eliminating engineering roles. They’re eliminating busywork.

So why does the narrative persist? Because it’s easier to write a headline about robots taking jobs than to explain what’s actually happening. And what’s actually happening is far more interesting, and far more consequential for anyone building or scaling a team.

The Jevons Paradox Is Playing Out in Real Time

In 1865, William Stanley Jevons observed something counterintuitive: when James Watt’s steam engine made coal more efficient to use, coal consumption didn’t decrease. It exploded. Because efficiency made new applications economically viable.

The same thing is happening with software engineering right now.

AI coding assistants make it 2-10x faster to produce working code. But that doesn’t mean companies need fewer engineers. It means thousands of projects that were previously too expensive to justify are suddenly viable. Internal tools that never got built. Automations that were “nice to have.” Products that couldn’t find margin with a six-month dev cycle but work perfectly with a six-week one.

I’ve seen this firsthand with our clients. A startup that budgeted for a bare-bones MVP is now getting a production-ready platform with monitoring, CI/CD, and proper test coverage, in the same timeline, for the same money. The output expanded. The headcount didn’t shrink.

The bottleneck has shifted. And that shift is where most companies are getting caught out.

The Bottleneck Was Never Code Production

Here’s the uncomfortable truth I keep running into: the teams struggling most with AI aren’t the ones writing bad code. They’re the ones with no engineering leadership.

When AI accelerates code production, everything downstream gets faster too, including the consequences of bad architecture, poor hiring decisions, missing processes, and undefined technical direction. You can ship a poorly architected system in half the time now. Congratulations. You’ll also need to rewrite it twice as fast.

The teams that are actually thriving with AI-assisted development share common traits, and none of them are about the tools:

• Clear technical direction. Someone decided what to build and why, before anyone opened a code editor.
• Defined engineering standards. Code review culture, testing expectations, deployment practices, the boring stuff that compounds.
• Hiring discipline. AI makes a good engineer great. It also makes a mediocre engineer dangerous. Someone needs to know the difference.
• Process that ships. Agile that actually works, not agile theatre. Incident response that’s rehearsed, not improvised.

These aren’t things AI provides. They’re things an engineering leader provides. And the faster your team can produce code, the more critical that leadership becomes.

What I’m Seeing on the Ground

Over the past year, I’ve worked with companies across the spectrum, funded startups, mid-market scale-ups, and enterprise teams running modernisation programmes. The pattern is remarkably consistent.

The companies with strong engineering leadership are using AI to amplify their existing culture. Their engineers write better code faster. Their architecture decisions hold up under increased velocity. Their processes absorb the acceleration without breaking. They’re shipping more, with higher quality, and their teams are happier because the boring parts of the job are disappearing.

The companies without it are drowning. More code is being produced, but nobody’s reviewing it properly. Architecture decisions are being made by default, not by design. The team is moving fast in four different directions simultaneously. Technical debt is accumulating at the same accelerated rate as features. And the founders are confused because they thought AI was supposed to make everything easier.

Sound familiar?

The Gap That AI Can’t Fill

AI is exceptional at generating code, explaining patterns, debugging, writing tests, and scaffolding infrastructure. I use it daily and it’s genuinely transformative.

But here’s what it cannot do:

• Set technical direction that aligns with where the business needs to be in 18 months, not just what the next sprint requires
• Build engineering culture, the norms, values, and practices that determine whether your team attracts and retains great people
• Make hiring decisions that balance technical skill with team dynamics, growth potential, and cultural contribution
• Navigate organisational politics to ensure engineering has the resources, autonomy, and air cover it needs to deliver
• Mentor engineers through the messy, human process of growing from competent to excellent
• Know when to say no, to the feature that sounds simple but will destroy your architecture, to the shortcut that’ll cost you six months later, to the hire that interviews well but won’t work in practice

These are fundamentally human judgments that require experience, empathy, and context. They’re the job of an engineering leader. And they’ve never been more important than right now, when the cost of getting them wrong is amplified by the speed at which code can be produced.

What This Means for Growing Companies

If you’re a startup that’s just raised a round and you’re about to scale your engineering team, you need leadership before you need more engineers. The decisions you make now about architecture, hiring standards, development practices, and technical culture will compound for years. AI won’t make those decisions for you. It’ll just make the consequences arrive faster.

If you’re a mid-market company whose tech lead just left, the gap isn’t “someone to write code.” It’s someone to set direction, maintain standards, and ensure the team doesn’t fragment under the pressure of AI-accelerated delivery. Every week without that leadership is a week of decisions being made by default.

If you’re an enterprise running a modernisation programme, AI-assisted development can dramatically accelerate your migration. But only if someone is making sound architectural decisions about what to migrate, in what order, and to what target state. Speed without direction is just chaos with better tooling.

The Bottom Line

The headlines will keep coming. AI will keep getting more capable. Engineers will keep being in demand, probably more so, not less.

But the companies that win won’t be the ones that adopt the most AI tools or generate the most code. They’ll be the ones with the engineering leadership to channel that increased capability into coherent, well-architected systems built by high-performing teams.

AI is raising the floor of what’s possible. Engineering leadership is what raises the ceiling.

If your team is producing more code than ever but you’re not sure it’s heading in the right direction, that’s not a tooling problem. That’s a leadership gap. And it’s worth addressing before the velocity makes it harder to course-correct.

The real question isn’t whether AI will replace your engineers. It’s whether you have the engineering leadership to make the most of what AI enables. If you do, the next few years will be extraordinary. If you don’t, all that speed is just going to get you to the wrong destination faster.