Code Without Coders, and the Economic Ripples of AI’s Software Revolution

The ATM didn’t eliminate bank tellers — it transformed banking into a more profitable industry focused on relationship building and wealth management. We’re approaching a similar inflection point with AI-powered software development, where scaling through compute — rather than headcount — will unlock new possibilities for business growth.

The economics are compelling: imagine a supermarket that can now afford to test thousands of checkout flow variations because the cost of implementation has dropped so dramatically. Projects that might yield just a 0.1% revenue increase — previously not worth a development team’s time — suddenly become viable when you can deploy AI agents at scale.

Reimagining Project Scope

The transformation will be particularly dramatic in how companies approach software development projects. Today’s model of assigning development teams to carefully prioritized projects based on resource constraints will give way to a more experimental, data-driven approach. When the cost of implementation plummets, companies can pursue incremental optimizations and ambitious moonshots simultaneously.

A perfect example of a moonshot is when we recently worked with a company struggling with Fortran code from 1962, undocumented and written by developers who have been six feet under for decades. Previously, modernizing such systems would have required massive teams and years of effort. Now, AI can analyze these ancient codebases, synthesize understanding from the code itself, and effectively create a “perfect employee” who knows every line of this legacy code. What was once an insurmountable technical debt becomes a tractable problem when you can scale through computing rather than headcount.

Imagine how this can change A/B testing in e-commerce in a slightly more modern example. Instead of testing two or three checkout flows, companies can experiment with thousands of variations, each tailored to specific user segments. You could optimize basket size, retention, or conversion rates across countless permutations — something impractical with traditional development resources.

The Evolution of Development Roles

This shift is already reshaping the software development profession. Our engineering team recently banned traditional development environments entirely and switched to pure AI-assisted development. The result? Our AI system became our most productive team member, merging more pull requests than any human engineer. But more importantly, it transformed how our engineers work. The traditional role of churning out code based on detailed specifications is becoming less relevant. Instead, we’re seeing a polarisation of the industry. On one end, there’s increasing demand for technical product managers who can precisely articulate requirements and understand system architectures. Conversely, we need engineers who can think strategically about implementation and review AI-generated code for long-term maintainability.

Eliminating technical debt becomes a realistic possibility when you can deploy AI agents around the clock. This means companies can focus more resources on revenue-generating initiatives rather than maintenance. When your foundations are stronger, you can build more ambitious features without worrying about shaky infrastructure.

However, this transition raises essential questions about liability and regulation. As we move toward an AI-generated codex, determining responsibility for failures becomes more complete. Is it a flaw in the pre-training? An implementation issue? A fine-tuning problem? The industry needs clear frameworks for liability, mainly when dealing with mission-critical systems.

From Trade to Creative Engine

Software development is evolving from a trade into a creative profession. Those who can orchestrate AI systems and think strategically about product development will be in high demand. Just as ATMs freed bank employees to focus on higher-value services, AI development tools will enable engineers to tackle more ambitious projects. At the same time, automated systems handle routine optimization and maintenance.

The companies that thrive in this new environment won’t be those who simply use AI to cut costs. Success will come to those who reimagine what’s possible when development resources are no longer the constraining factor. In this scenario, human creativity and strategic thinking become more important than ever.

The Next Economic Leap

The economic implications extend beyond individual companies to entire regions and economies. Those with lower regulatory friction will likely see faster adoption and more innovation. Just as some areas skipped the PC era and went straight to mobile, developing economies might leap directly into AI-first development, bypassing traditional software development models entirely.

AI will change the economics of what can be built, and having the best ideas about what to build will make it the most sought-after differentiator.