Why We're Doubling Down on Developers While Everyone Else Is Betting on AI

Every week, another headline declares developers obsolete. AI can write code now. Natural language is the new programming language. The 10x engineer is dead - long live the 100x prompt.

And every week, I see another project go sideways. Broken. Behind schedule. Hemorrhaging money. Built by teams who believed the hype.

We're not buying it. In fact, we're making the opposite bet.

The Jevons Paradox Nobody Wants to Talk About

In 1865, economist William Jevons observed something counterintuitive: as steam engines became more efficient at burning coal, total coal consumption went up, not down. More efficiency meant more applications, more scale, more demand.

The same thing is happening with software.

AI tools are making code easier to generate. And you know what that means? More software. More features. More integrations. More complexity. The backlog isn't shrinking, it's exploding. Every company is suddenly a software company. Every department wants an app. Every process needs automation.

The typing was never the bottleneck. We're not running out of keystrokes. We're running out of people who can think clearly about what to build, why to build it, and how to build it so it doesn't collapse under its own weight six months from now.

The Hard Part Was Never Typing

Here's what AI is genuinely good at: generating code that looks correct. Autocompleting patterns. Scaffolding boilerplate. Moving fast on the parts that were already figured out.

Here's what breaks projects:

• Nobody defined what "done" actually means
• The architecture doesn't account for how the business will evolve
• Three services are doing the same thing slightly differently
• The team optimized for shipping, not for maintaining
• Edge cases got handwaved because they were "rare"
• Someone copy-pasted a solution without understanding why it worked

AI doesn't solve any of this. AI makes it worse. You can generate bad decisions faster than ever. You can ship technical debt at unprecedented velocity. You can build yourself into a corner with incredible efficiency.

I've seen the wreckage. Teams drowning after their AI-assisted sprint turned into an AI-assisted disaster. The code exists. Mountains of it. But nobody knows what it does, why it does it, or how to change it without breaking everything else.

Power Tool vs. Replacement

A nail gun doesn't replace a carpenter. It makes a good carpenter faster. Give it to someone who doesn't understand framing, and you just get a dangerous mess, quicker.

That's AI in software development. It's a power tool. In the hands of an experienced developer who understands systems thinking, trade-offs, and maintainability, it's incredible. In the hands of someone who thinks software is just text that compiles, it's a liability.

The companies cutting their engineering teams and betting on AI are making a category error. They see code as the product. It's not. Code is a byproduct of thousands of decisions, each one shaped by context, constraints, and consequences that compound over time.

Good developers don't just write code. They prevent bad code. They ask the questions that save you six months. They see the dependency that's going to bite you in production. They know when the "quick fix" is actually a trap.

AI can't do that. AI doesn't know your business. AI doesn't remember why you made that weird architectural choice two years ago (it was because the vendor API had an undocumented rate limit that only showed up under load). AI doesn't push back when the requirements don't make sense.

Our Bet

So while the industry races to replace developers, we're investing in them. Hiring them. Training them. Building a culture where thinking is valued as much as shipping.

We're not anti-AI. Our team uses these tools daily. But we use them the way a surgeon uses a robot arm - to be more precise, not to skip medical school.

When projects go sideways, it's almost never because the developers couldn't type fast enough. It's because somewhere along the way, understanding got sacrificed for velocity. Context got lost. Someone assumed the tool was smarter than it is.

We're betting that as AI makes code generation trivial, the premium on developers who can think - who can architect, debug, communicate, and make judgment calls under uncertainty - is going to go through the roof.

We're betting that the companies who survive the next decade won't be the ones who cut their engineering teams first. They'll be the ones who figured out how to combine human judgment with AI leverage.

We're betting on developers. Not because we're nostalgic. Because we've seen what happens when you don't.

Ephesus Gates is a software product development firm that helps companies build products that last. We bring experienced developers who know how to think, not just ship. If your project needs that kind of help, let's talk.