profile

Software Architecture Insights and AI/ligned

Software Architecture Insights offers practical thinking on architecture, cloud scale, and technical leadership, written by Lee Atchison, a working CTO and software architect. Every issue draws on decades of real production experience, not theory. Subscribe for insight you can put to work right away. AI/ligned offers AI ethics for technical leaders, not policy theorists. Written for the people who actually build and deploy AI systems, AI/ligned covers what responsible AI looks like when you are the one accountable for the outcome. Practical, not academic.

The Move From Execution To Origination

This article is a reprint of the article I wrote last week on The Software Conductor substack. Let me give you the numbers straight, because I think the industry has done a lot of hand-waving around them and that isn’t helping anyone. Programmer employment in the U.S. fell more than 27% over the past two years. Entry-level technology hiring at the largest firms is down 25% year over year. These aren’t rounding errors or statistical noise. They represent hundreds of thousands of people who...

The Line AI Is Drawing in Your Architecture Team

You've probably read a dozen articles about how AI tools are changing software development. Code generation. Developer productivity. Faster PRs, smaller backlogs, fewer boilerplate hours. Most of this coverage focuses on what developers do with their time, which is genuinely interesting. But there's a different question that matters more for your architecture practice. What happens to the architecture team when AI absorbs the execution layer of software work? That question has a different...

Why I Wrote The Software Conductor

I started writing this book three times. The first two attempts were what I thought a software architecture book was supposed to be. Frameworks. Diagrams. Bulleted lists explaining the difference between a developer and an architect. Sections with titles like "Characteristics of Effective Architecture Practice." The book was technically correct. It was also completely lifeless. I called the publisher and canceled the contract. That was three years ago. The reason it was lifeless took me a...

Working as Intended

There’s a phrase you’ve probably typed into a ticket at some point in your career. Maybe someone filed a bug. Maybe a customer complained. You looked at the system, traced the behavior, and confirmed that the output was exactly what the logic should produce. So you closed it: Working as intended. It’s the cleanest possible resolution. It means the system did what the system was supposed to do. It means the problem, if there is one, belongs to someone else: product, policy, legal, whoever...

Prompt Injection Isn’t a Bug. It’s a Property

If you worked on web applications fifteen or twenty years ago, you remember SQL injection. You'd be doing a code review, spot a line where the user's input was being concatenated straight into a SQL query, and feel a small chill. You knew what could happen next. You'd file a bug, talk to the developer, and replace the concatenation with a parameterized query. Over time, the industry made SQL injection hard to get wrong. Frameworks handled queries for you. ORMs treated raw SQL as a thing to...

When the Bill Comes Due: Amazon's AI Bet and the Engineers It Let Go

Amazon experienced something that should give every technology leader pause. The company recently suffered multiple high-severity production incidents. These events were serious enough that Amazon's SVP of eCommerce Foundation, Dave Treadwell, convened a meeting of retail technology leaders to perform a "deep dive" into what went wrong. His internal memo described the incidents as having a "high blast radius." At least one was linked to AI-assisted coding changes.At roughly the same time,...

AI Promised 10x Developers. The Reality Is Far More Complicated

The promise of AI coding tools has been nothing short of astonishing. AI will replace your need to hire software developers, and transform your existing developers into 10x engineers. According to this commonly believed view, AI will revolutionize software development productivity across the board, making the entire world better. But is this truly a potential reality? A new report from GitClear paints a very different picture. After studying over 2,000 developer-weeks of data, the report...