The Rise of the Artisan-Builder: Software Engineering in the AI Era

Artisans who are builders and builders who are artisans

Google CEO recently reported a massive shift in internal development workflows. As of April 2026, 75% of all new code at Google is AI-generated.

Why This Matters

There is a widening gap between ‘vibe-coding’—the ability to rapidly prototype using LLMs—and the engineering rigor required for production-grade systems. While AI can generate thousands of lines of code daily, this often results in insecure, broken ‘code spaghetti’ that lacks scalability. The technical reality is that while prototypes are easy to ship, rock-solid platforms (such as banking systems) require deep architectural expertise to prevent catastrophic failures and security vulnerabilities that non-technical builders cannot foresee.

Key Insights

• AI adoption is widespread among professionals, with 84% of developers and technologists adopting AI as of the 2025 Developer Survey.
• The ‘Artisan vs. Builder’ dichotomy distinguishes between those who prioritize speed/efficiency (IKEA-style assembly) and those with the craft knowledge to ensure structural integrity and security.
• Developer identity is shifting from ‘those who can write code’ to those who possess ‘taste’—the expertise to know what to build and how to guide an LLM to execute it correctly.
• Technical expertise remains indispensable for ‘base-of-the-pyramid’ software like .NET or Windows, where acute utility is insufficient compared to the need for secure, scalable foundations.

Practical Applications

• । Use case: Rapid prototyping for early-stage applications using AI coding assistants to increase delivery speed.
• । Pitfall: Relying on non-technical ‘vibe coding’ for core infrastructure, leading to hard-coded errors and unscalable architectures.

References:

• https://stackoverflow.blog/2026/05/28/artisans-and-builders/