Every programmer will be fullstack

Just a few years ago, if you wanted to ship production-grade software, you had to raise a small army.

You'd recruit a frontend engineer to build React components and optimize loading times, a backend engineer to design APIs and manage database queries, a DevOps engineer to set up CI/CD pipelines and manage Kubernetes clusters, a site reliability engineer to monitor system health and respond to outages, a platform engineer to maintain internal infra, and a whole platoon of other specialists.

This specialization emerged naturally as software grew more complex and demanding. Frontend development became its own discipline when browsers became a capable new runtime environment. Backend specialization followed as server-side development demanded expertise in databases, API design, security, and performance optimization—skills that had little overlap with frontend work. The DevOps role crystallized in the late 2000s when companies needed to deploy code multiple times per day while maintaining 99.9% uptime, requiring specialists who could automate deployments and build resilient infrastructure.

Today, AI tools make every programmer fullstack by default and a specialist by exception.

Codegen tools have contributed to this shift towards default fullstack engineers. They remove the syntax barriers that foster specialization. A good Python developer understands system design and composition, but they may not fully grasp the nuances of React hooks. So they'd struggle to build a basic full web app. Today, that programmer can spec and create a complete web app with the help of Cursor or Claude Code.

Another trend contributing to the shift towards fullstack programmers is new tools that productize roles into software. As AI models improve at reasoning through code, entire specialized functions are productized into AI-powered tools. 

Consider AI SRE tools like Cleric and Traversal, which scan logs, identify patterns, and execute playbooks that would have taken senior SREs years to master. Or AI Pentesting Tools like XBow or Sybil, which probe live code with the same chain-of-attack graphs a red team would craft by hand. The heavy lifting is still happening, it's just wrapped in an API that a single fullstack dev can call.

I'm deeply interested in the tools that are accelerating this trend and reshaping the identity of a software engineer. 

The old metaphor for a highly capable developer was the T-shaped engineer: possessing deep expertise in one domain, with broad, yet shallow, knowledge across others. But that T-shape is rapidly growing new stems of deep, AI-assisted capability. We're moving toward a model that resembles less a 'T' and more a 'Comb' — multiple prongs of specialized power available to a single individual.