AI Is Changing the Landscape.
But Probably Not in the Way You Think.
The interesting question isn't whether AI will replace developers. It's what happens when the cost of building software falls, the speed of experimentation increases, and businesses suddenly have far more capability than they know what to do with.
Every few years, technology gives everyone the same uncomfortable feeling:
“Things have just changed.”
The web did it.
Cloud computing did it.
Smartphones did it.
And AI is doing it again.
Except this time the change is happening unusually quickly.
AI can write code. It can analyse documents. It can generate images. It can summarise meetings. It can query data. It can explain an unfamiliar codebase. It can produce a first draft of something that previously took a person an afternoon.
That creates an obvious question:
“If AI can do all of that, what happens to the people who used to do it?”
I think that's the wrong question.
The more interesting question is:
“What becomes possible when the cost of doing those things drops dramatically?”
The first thing AI changes is the cost of experimentation
For years, one of the biggest constraints in software wasn't whether something was technically possible.
It was whether it was worth spending the time to find out.
Someone has an idea.
It needs a developer.
The developer needs to understand the requirement.
They need to design it. Build it. Test it. Deploy it. Maintain it.
Suddenly a two-day experiment has become a two-week project.
AI changes that equation.
A developer can use AI to explore an unfamiliar library, produce boilerplate, write tests, transform data, generate prototypes and investigate alternatives much faster.
That doesn't mean the developer has disappeared.
It means the price of asking “what if?” has fallen.
That creates a strange new problem: too many possibilities
For a long time, businesses had the opposite problem.
They had more ideas than they had engineering capacity.
AI starts to narrow that gap.
And that sounds entirely positive until you realise what happens next.
People start building things simply because they can.
Ten internal tools.
Five dashboards.
Three automations.
A chatbot.
An AI assistant.
Another AI assistant because the first one didn't have enough branding.
Now you have the opposite problem.
The constraint moves from “Can we build it?” to “Should we build it?”
That's a much more interesting problem for technology leaders.
AI doesn't remove the need for architecture
There is a temptation to think that AI makes architecture less important.
I think it makes good architecture more important.
If AI makes it easier to produce code, the amount of code an organisation can produce increases.
But code is not the same thing as a system.
You can generate an API in seconds.
You still need to decide:
What data should it access?
Who is allowed to use it?
What happens when it fails?
What is the source of truth?
How do we monitor it?
How does it fit into the existing architecture?
What happens six months from now?
AI is exceptionally good at producing components.
Businesses still need people who understand the system those components belong to.
The value of knowing how to code is changing
This is probably the part that makes developers most uncomfortable.
I don't think knowing how to write software becomes worthless.
I think the value shifts.
Syntax becomes less important.
Judgement becomes more important.
Knowing how to ask an AI to write a function is useful.
Knowing whether the function is the right solution is much more useful.
Knowing how to debug a generated solution is useful.
Knowing whether the architecture that produced the problem should exist in the first place is better.
This is one reason I'm interested in the space between technology and business.
I've spent enough years writing software to know that writing the software is often the easy part.
The specification becomes a superpower
There is another consequence of AI that doesn't get enough attention.
The better you can describe what you want, the better AI can help you build it.
That makes requirements more valuable.
Not long documents full of corporate language.
Clear thinking.
What problem are we solving?
Who has the problem?
What does success look like?
What data do we need?
What constraints matter?
What must never happen?
How will we know it works?
Those questions are becoming the interface between humans and increasingly capable machines.
The better the thinking going in, the more useful the machine becomes.
This changes product management too
Product teams have traditionally had to make trade-offs based partly on engineering capacity.
“We could build that, but it would take three months.”
What happens when it might take three days?
The backlog changes.
The economics change.
The definition of an MVP changes.
And potentially, the organisational structure changes.
If small teams can accomplish much more with AI-assisted development, the advantage may move away from having enormous delivery capacity and toward having excellent product judgement.
That doesn't mean every company needs fewer people.
It means the output expected from each person can increase.
AI also changes the meaning of “technical”
Historically, there was a fairly clear divide.
Technical people built things.
Business people decided what things should be built.
That boundary was never completely real, but AI makes it even harder to maintain.
A product manager can prototype.
A marketer can analyse data.
An operations manager can automate a workflow.
A founder can build an internal tool.
A developer can spend more time understanding the business instead of translating requirements into repetitive code.
That is potentially a very good thing.
But it also means organisations need people who can connect those capabilities together.
Where I think the real opportunity is
I'm less interested in building an AI wrapper around something just because it has become fashionable.
The interesting opportunities are where AI can sit inside an existing business process and create leverage.
Imagine an e-commerce operation with thousands of products.
AI could help classify products, detect anomalies, improve descriptions, identify missing attributes, analyse customer questions, spot unusual orders or summarise performance.
But the value isn't “we used AI.”
The value is that somebody previously spent twenty hours doing a repetitive task and now spends two hours reviewing the output.
Or that a problem is detected before a customer notices it.
Or that a team can test ten ideas instead of two.
That's the measure I care about.
AI doesn't eliminate bad processes
This is another trap.
If a business process is fundamentally broken, adding AI to it doesn't magically fix the process.
You can automate a bad workflow.
You can make a bad decision faster.
You can generate beautiful explanations for completely unreliable data.
In other words:
AI can increase the speed of a system. It can also increase the speed of a mistake.
That's why the fundamentals still matter.
Good data.
Clear ownership.
Defined processes.
Good architecture.
Security.
Testing.
Human accountability.
AI sits on top of those things.
The new competitive advantage might be organisational, not technical
This is the part I find most exciting.
Two companies can have access to exactly the same AI models.
One does very little with them.
The other redesigns how its people work.
It connects internal data. Removes repetitive work. Gives employees better tools. Speeds up experimentation. Improves customer service. Builds faster feedback loops.
Same underlying technology.
Completely different outcome.
That suggests the competitive advantage isn't necessarily the model.
It's the organisation built around it.
And this is where my own role is changing
I've spent roughly 18 years working across software, e-commerce, infrastructure, marketing, integrations and systems.
For a long time, being able to build the thing was a huge advantage.
It still is.
But AI means more people can now get surprisingly far without spending years learning how to build every component themselves.
So the differentiator becomes increasingly about what happens before and after the code.
Understanding the business.
Finding the bottleneck.
Choosing the right problem.
Designing the system.
Knowing what should be automated.
Knowing what should stay human.
Measuring whether it actually worked.
And having enough technical depth to know when an AI-generated answer is brilliant, plausible nonsense, or somewhere in between.
The landscape is changing. That's the point.
I don't think the useful question is whether AI is going to replace developers, product managers, analysts, marketers or designers.
Some individual tasks will absolutely disappear.
Some roles will change dramatically.
New roles will appear.
Some companies will use AI incredibly well.
Others will spend two years adding chatbots to things that didn't need chatbots.
That's normal technological change.
The people who benefit most won't necessarily be the people who know the most about AI.
They'll be the people who understand where AI creates leverage.
That's a subtle but important distinction.
AI isn't the strategy. It's another component of the system.
The interesting work is figuring out where that component belongs.
And I suspect that's going to be one of the defining technology questions of the next decade.