The future of software engineering might look like... plumbing.

Patrick Mathieu

Patrick Mathieu

— 3 min read
Share
The future of software engineering might look like... plumbing.

Hey all,

Lots of things happened this year, but I did not forget about the blog, even though I did not put enough effort into it.

This blog is obviously an information security blog, but I think it's always important to also add the human side and a little personal touch.

After leaving DoorDash almost 500 days (16 months) ago, I took a lot of time for myself and enjoyed the summer. And at the end of 2025, I went full-time consulting with my own company, Hexius Technologies, and I built a SaaS based on a system I wrote in Python a few years ago: an attack surface analysis SaaS (Attack Surface Management, ASM) called Hexposure (OK no more shameless plugs), but with the objective of finding more issues than the competitors and making it more useful (not just a list of vulnerabilities, like 90%+ of the vendors out there are doing).

So yes, only 3 blog posted happened, but many clicks are still happening on many of the articles, showing the need to continue writing 😄

But now, obviously we have to discuss AI a lot more... (the statistical, non-intelligent machine learning thing everybody uses today).

After 1.5 years of almost full-time architecting, "coding," and speeding up my development with Claude AI, I also started helping startups build PoCs for their systems with the help of AI. I've followed the creators of Claude and other AIs as much as I can for tips and tricks on how to use them optimally, etc.

And the more time passes, the better the code gets, and the more obvious the future becomes...

Story time

It's 2035, you run a company, and your application is broken. Someone hacked it, or is abusing it. Something is definitely wrong.

The AI that wrote it says everything looks great. (Let's dream that these won't be perfect by then.)

So you call the code plumber. He opens an actual file. Reads it. Line by line. Like it's 1999 (or 2019). Your team stands around in awe. "He… understands the code?" 😮

Two hours later, it's fixed. The invoice says:

  • Diagnostic: $2,500
  • Knowing which keys to type to fix it: $9,500

Tomorrow's programmers won't be the ones writing all the code, but they might be the rare breed who still understand it, and who get called at midnight.

The developer role is dead. Long live the developer role? ... Time to learn to read code again, I guess!

Outro

The way things are going, I don't think anybody is really going to lose their job, but we will definitely have to adapt, just as horse carriage drivers changed to car delivery over time. It's not a net-new job, but a transformation.

Today's AI developers (vibe coders, whatever we want to call them) go faster and are great at what they do if, and only if, they are great at what they do without AI.

Anyone can vibe code an app, but it will be a proof of concept: it won't scale or be ready for production, nor will they know how, or have the skills and time, to deploy an app across a cloud, a swarm of Docker containers, etc. But a DevOps engineer who is an expert at it can, and AI definitely helps them go faster and do more, even a bit outside their skill set, because they still understand the concepts and how it works under the machine.

So I think understanding code will definitely be important in the future, whether we code with AI or not.

What do you think? Where are we going, and what is important to keep learning today as a senior in the domain? Or, more importantly, what should juniors learn today to be ready for tomorrow?

Read more