My Take on AI in Software Development

November 18, 2026
By Armen Stein
My Take on AI in Software Development

AI is causing some pretty big waves in the software development industry, and we’re riding them, too. At J Street, we’re using AI for all kinds of things: understanding meetings, writing software specifications, designing databases, coding, and automated testing.

We recently completed a mammoth data migration project in under 20 hours. Four other companies told the customer it couldn’t be done. In a way, they were right: Without AI, the project was not feasible.

Better Software in Less Time

I’ve already seen AI reduce the hours we spend on projects while helping us improve quality. I’m interested in where that leads, because I think cheaper, faster software development will open up a lot of opportunities for businesses that haven’t been able to afford it before.

There’s a theory called the Jevons Paradox that I think applies here. The idea is that when something becomes more efficient and cheaper, people will use more of it. We could see the same thing happening with software. As our costs come down, more businesses may decide it’s worth investing in custom software. I’ve talked to plenty of businesses that have been making do with spreadsheets or clunky old systems because the cost of replacing them didn’t make sense. Maybe now it will.

But Vibe Coding Isn’t It

You may have heard about vibe coding. If you’re not familiar with the term, it’s using AI to build software by describing what you want, rather than writing the code yourself. But here’s the important part: the AI is writing code that you don’t know how to write. You can get some impressive initial results. But there’s a catch. Getting something to work once is very different from building something a business can rely on every day for years to come.

A lot of the hard work in software development happens before anyone writes a line of code. You must understand the business, how people do their jobs, and what they’re trying to accomplish. You must ask the questions they may not have thought of asking themselves. What happens when the business grows? Will the software scale? Where are things likely to slow down? Is the system secure? Will anyone be able to maintain the software six months or five years from now?

Then, there’s the human factor. Customers don’t always know exactly what they want until they see it. Sometimes two requirements conflict with each other. Sometimes a conversation reveals that the original idea wasn’t really the right solution. Working through all of that takes experience and judgment.

I think of AI like Jarvis from Iron Man. Jarvis is incredibly capable, but Tony Stark is still the one making decisions in the lab and guiding what happens inside the suit. I’d be hesitant to send Jarvis out on his own without Tony keeping an eye on things.

For every story about an impressive system built with AI, there’s another about a project that burned through a pile of AI tokens and electricity only to produce software that barely works, has security holes, can’t handle real-world demand, or is so convoluted that nobody can maintain it.

You can generate a lot of code in a hurry with AI. That doesn’t mean you’ve built a good system. Some companies are starting to ease off the AI throttle after pushing hard to adopt it. I’m not surprised. There’s a lot of potential here, but using AI well takes more thought than handing it a prompt and hoping for the best.

What This Means for J Street

We’ve always had more business analysts and software designers than programmers at J Street – and that’s intentional. We spend a lot of time understanding what customers need and figuring out how to turn that into something that works in the real world. And, as anyone who’s worked on business software knows, that requires detailed conversations between humans.

We’re using AI at every stage of our work and learning more everyday. We need to. If we don’t adopt these tools, we’ll quickly be outmatched by companies that do. But I don’t believe the winning strategy is to replace human expertise with AI. It's to combine the two.

I don’t see AI eliminating the need for what we do. If anything, it makes our experience more valuable. We can use AI to do things faster and take on work that might not have made financial sense before. We still need people who know which questions to ask, when a proposed solution doesn’t add up, and what could go wrong down the road.

This is where I see my role, and J Street’s, continuing to evolve.

I don’t know exactly where all this will lead. Technology is moving too quickly to make confident predictions about that. But I’m excited about what we’re already able to accomplish, and I’m looking forward to seeing what else we can do with it.

 

As always,
Armen

Interested in talking more about how we approach this? We’d love to. Just Schedule a free consultation or give us a call at 425-679-6206.

‍