AI: a complete paradigm shift, or just another dev tool?

This post was originally published on my Linkedin.

AI: a complete paradigm shift, or just another dev tool? This duality isn’t a contradiction; it’s just the reality of how transformative technologies integrate into existing systems and workflows.

I’ve been thinking a lot about the similarities between today’s AI discussions and the conversations we had around containerization back in 2013/2014, when Docker came onto the scene.

AI as a paradigm shift

With AI, we’re changing not just what we build, but how we can build it. The day-to-day life of development has been deeply altered by AI-assisted coding.

In terms of what we can build, AI enables capabilities that were previously impossible or impractical. We’re building products that leverage things like natural language processing in ways that solve existing problems in entirely new ways – while also creating new classes of problems.

The technology has opened up solution spaces that didn’t exist before, much like how containers fundamentally changed our approach to building, shipping, and running applications, not to mention the control plane layer on top of it all.

AI as “just another dev tool”

However, adopting AI reveals many of the same organizational and technical challenges we encountered with containerization. When companies first implemented container strategies, they often discovered that containerization exposed existing architectural problems and highlighted inefficiencies in their software development lifecycle that had been masked by previous approaches.

AI is following a similar pattern. While we can generate code significantly faster than before, the speed of code generation doesn’t automatically translate to higher quality software. The fundamental challenges of code review, architecture, security, and maintainability are still there (though we can use AI to solve some of those problems now, too). In many cases, the increased velocity has simply moved bottlenecks to other parts of the development process.

Teams that expected AI to eliminate technical debt or bypass established engineering practices have found that these foundational issues still require direct attention. The technology amplifies existing strengths and weaknesses – but also amplifies the bad stuff.

Looking forward

Organizations that invest in understanding appropriate use cases, provide adequate training, and maintain realistic expectations about outcomes tend to achieve better results than those expecting immediate transformation.

Neither containers nor AI function as universal solutions. They excel in specific contexts and require thoughtful integration with existing systems and practices. The value comes from strategic application rather than broad adoption.

The companies positioning themselves well to get the most out of AI are those that approach it with the same rigor they apply to any significant technology decision. They’re identifying specific problems AI can solve effectively, building the necessary organizational capabilities, measuring impact (like with the AI Measurement Framework) and maintaining focus on fundamental software engineering principles.