Software fundamentals matter more in the age of AI coding tools
In a talk at AI Engineer, Matt Pocock argues that developers who succeed with AI coding agents are those who ground their work in decades-old engineering fundamentals, not those who delegate everything to AI.
Score breakdown
Developers adopting AI coding agents should audit their engineering practices first — Pocock's framework suggests that fundamentals like TDD and vertical slices are the leverage point that separates high-quality AI-assisted output from unmaintainable code.
- 01Matt Pocock (@mattpocockuk) delivered this talk at the AI Engineer conference.
- 02He draws on 18 months of teaching developers to build with AI agents.
- 03His thesis: AI coding tools can produce spaghetti code faster than any human team when used poorly.
Matt Pocock's talk at AI Engineer challenges the framing that AI coding tools are simply a productivity multiplier anyone can tap into equally. Based on 18 months of teaching developers to build applications with AI agent swarms, he argues that the developers who consistently ship high-quality software are those who apply foundational engineering principles — not those who hand off the most work to AI or those who refuse to use it at all.
The specific fundamentals Pocock highlights — ubiquitous language, vertical slices, test-driven development (TDD), and deep modules — are well-established concepts that predate the current AI wave by decades.
The specific fundamentals Pocock highlights — ubiquitous language, vertical slices, test-driven development (TDD), and deep modules — are well-established concepts that predate the current AI wave by decades. His central thesis is that these principles didn't become obsolete with the rise of AI coding agents; rather, they became more critical, because AI tools can generate spaghetti code at a pace no human team could match when used without disciplined process. The talk presents an iterative process his students use to ship with AI agent swarms while keeping codebases maintainable.
Key facts
- 01Matt Pocock (@mattpocockuk) delivered this talk at the AI Engineer conference.
- 02He draws on 18 months of teaching developers to build with AI agents.
- 03His thesis: AI coding tools can produce spaghetti code faster than any human team when used poorly.
- 04Successful developers are those who neither delegate everything nor nothing to AI.
- 05The engineering fundamentals he highlights are ubiquitous language, vertical slices, TDD, and deep modules.
- 06These are decades-old principles that Pocock argues have grown more important, not less, in the AI era.