When the Tools Get Smarter

How LLMs are reshaping the future of software engineering.

Written February 2026, when the new generation of LLMs had just become widely available. This is the earliest piece in the collection: a software-engineering frame on the same questions the later essays return to in more literary register.

The rise of large language models has triggered a familiar anxiety across the software world: if AI can write code, what happens to the engineers? The fear is understandable. For decades, software engineering has been a high-skill, high-demand profession. Now, suddenly, the tools themselves can perform much of the work that once required years of training.

But the deeper story is not one of replacement. It is one of redistribution: of skills, of leverage, and of opportunity.

This moment is not a collapse. It is a shift in where value sits.

The new leverage: when individuals gain the power of teams

The most profound change is not that LLMs can write code. It's that they compress the cost of creation. A single engineer can now:

• prototype in days instead of months
• build full-stack applications without deep specialisation
• explore ten ideas instead of one
• operate without the overhead of a large team

This is the same pattern we saw with cloud computing, open-source, and app stores. Each removed a barrier. Each expanded who could build. Each triggered a wave of innovation from unexpected places.

LLMs are simply the next compression layer — and the most powerful one yet.

The result is a surge in solo and micro-founder innovation. Engineers leaving medium and large companies don't stop building; they stop building for someone else. With AI as a force multiplier, they can create tools, products, and entire businesses that once required a small army.

This is not doom. It is a redistribution of creative power.

Why the world doesn't get swamped by infinite apps

If building becomes cheap, why doesn't the world drown in software?

Because the bottleneck has moved.

1. Distribution is still hard. Getting users, earning trust, and integrating into real workflows remain human challenges.

2. Complex systems still require teams. Enterprise-scale software still needs architecture, governance, security, and long-term stewardship.

3. Product–market fit remains the ultimate filter. LLMs can generate code, but they cannot generate demand. Most ideas still fail — and that's healthy.

The explosion in creation does not lead to chaos. It leads to more experiments, and therefore more black swans: rare, high-impact successes emerging from a larger pool of attempts.

The human pattern: how engineers actually respond to disruption

Across industries and eras, three behavioural patterns repeat when technology reshapes a profession:

• Most people seek stability, not risk. They stay in structured roles unless pushed.
• A minority goes independent, and this minority disproportionately shapes the future.
• Disruption increases variance, not uniformity: some fall behind, some adapt, a few leap ahead.

LLMs don't eliminate engineers. They eliminate average engineering.

What remains, and becomes more valuable, are the higher-order skills:

• system design
• domain expertise
• architectural thinking
• understanding constraints
• reasoning about trade-offs
• communicating intent

These are the skills that define senior engineers, and they are precisely the skills AI cannot automate.

A black swan moment — but not a catastrophic one

This shift has the flavour of a black swan: sudden, high-impact, and transformative. But it is not catastrophic. It is the logical next step in the digitisation curve.

What is unprecedented is the speed at which leverage increases. A single engineer with an LLM can now do what once required a 20-person team. That changes the economics of innovation, the structure of companies, and the trajectory of careers.

But it also opens the door to something new: a world where individuals with deep domain knowledge can build tools that previously required institutional backing.

This is empowerment, not erasure.

The future belongs to builders, not coders

The engineers who struggle will be those whose value was tied to routine implementation. The engineers who thrive will be those who can:

• frame problems
• design systems
• integrate components
• understand real-world constraints
• bring domain expertise to bear

LLMs don't replace engineers. They replace the parts of engineering that were never the true source of value.

The future belongs to those who can think, not just type.

For a longer response, send a reply →

← All essays