This is part 4 of our series on The Democratisation of Building.
Let’s talk about the elephant in the room. We’re a digital agency writing about how AI is dismantling the digital agency model. If that feels like turkeys voting for Christmas, good — it means we’re being honest.
Dries Buytaert, creator of Drupal and someone who’s watched the web industry evolve for two decades, recently compared AI’s impact to the Industrial Revolution. Not in that hand-wavy “this changes everything” way people trot out at conferences. In the specific, structural sense: entire categories of skilled labour are being fundamentally repriced. The loom didn’t eliminate textiles. It eliminated the old economics of textiles. AI is doing the same to digital services.
The Maths Stops Working
Here’s the core problem for every agency still billing by the hour: when AI does in minutes what used to take hours, hourly billing becomes indefensible.
Think about it from the client’s perspective. You commission a website redesign. The agency quotes 200 hours. But their developers are using AI tools that compress what was genuinely 200 hours of work into 40. Do you pay for 200 hours? 40 hours? Or do you start asking entirely different questions about what you’re actually buying?
This isn’t hypothetical. It’s happening now, across every discipline simultaneously. Content creation, code generation, UI design, data analysis – the compression is everywhere. A junior developer with Claude or Cursor can now produce work that would have required a senior developer two years ago. Not identical work, mind you (we’ll come back to that), but work that passes the “good enough” bar for a significant chunk of deliverables.
The traditional agency bundle – strategy, design, development, content, SEO, ongoing support – was held together by the fact that all of these things were expensive and time-consuming. When the time and cost collapse, the bundle falls apart.
That’s the unbundling.
What Gets Commoditised
Let’s be specific about what’s being commoditised, because vagueness helps no one.
Template-based design and development. If the brief is “build me a website that looks like this competitor but with our branding,” AI can do that. Increasingly well. The agencies that built their business on this kind of work are in trouble, and pretending otherwise is dishonest.
Content production at volume. Blog posts, product descriptions, social media copy, email sequences — the kind of content that agencies churned out at scale. AI produces this faster, cheaper, and (let’s be honest) at roughly the same quality as the average agency output.
Traditional SEO. This one’s a double hit. Not only can AI produce SEO-optimised content at scale, but the entire premise of SEO is shifting as AI intermediates search. When users get answers from ChatGPT instead of clicking through to your lovingly optimised blog post, the ROI calculation for traditional SEO collapses. Agencies that sold SEO retainers as their bread and butter need a new menu.
Routine development tasks. Bug fixes, feature implementations from well-defined specs, API integrations, migrations between well-documented platforms. These are pattern-matching exercises, and AI is very good at pattern matching.
What Survives
Here’s where it gets interesting – and where we stop being gloomy and start being honest about what actually matters.
The 80/20 split we talked about in the first post in this series applies here too. AI handles roughly 80% of the execution. But that remaining 20% is where all the value lives. And it turns out that 20% is what good agencies were always selling — it was just bundled with (and subsidised by) the 80% of commodity work.
Understanding the actual problem. Most projects fail not because the code was bad but because the wrong thing got built. The ability to sit with a client, interrogate their assumptions, understand their users, and define a problem worth solving — that’s human work. It requires empathy, experience, and the kind of lateral thinking that comes from having seen dozens of similar projects go wrong in different ways.
Taste. This is the word that keeps coming up in our internal conversations. AI can generate a hundred design options. It cannot tell you which one is right for this brand, this audience, this moment. Taste is accumulated judgment. It’s pattern recognition built on years of seeing what works and what doesn’t.
Quality assurance and governance. AI generates code and content at speed. It also generates bugs, security vulnerabilities, hallucinated facts, and architectural decisions that look fine today and become catastrophic at scale. Someone needs to catch that. And – this is the critical bit – that someone needs enough experience to know what to look for.
The grey-haired builders. We’ve used this phrase before and we’ll keep using it because it captures something important. In a world where anyone can build, the people who’ve been building for twenty years become more valuable, not less. Not because they type faster or know more keyboard shortcuts, but because they know what good looks like. They’ve seen the failure modes. They know why that “quick and easy” approach will cost you ten times more to fix later. Experience isn’t a nice-to-have in the age of AI. It’s the quality control layer that prevents AI-generated output from becoming AI-generated chaos.
The Agency That Survives
The agencies that thrive through this unbundling won’t be the ones that pretend AI isn’t happening. They won’t be the ones that quietly use AI while still billing hourly rates based on pre-AI timelines (clients aren’t stupid — they’ll figure it out). And they won’t be the ones that panic and slash their prices to compete with AI tools directly.
The agencies that survive will be the ones that explicitly separate their value from their labour. That price on outcomes — a working product, a measurable improvement, a problem solved – rather than on hours spent. That invest in the human capabilities AI can’t replicate: strategic thinking, quality judgment, domain expertise, and the experience to know where the landmines are buried.
This isn’t doom. It’s evolution. And honestly, it’s overdue. Too many agencies have been selling hours when they should have been selling outcomes. Too many have padded timesheets with work that didn’t need doing. The AI unbundling forces a reckoning, and that reckoning is healthy.
Where We Stand
We’d be hypocrites if we wrote all this without being transparent about our own position.
Versantus has always been outcomes-focused. We’ve never been comfortable with the pure time-and-materials model, because it creates perverse incentives – the longer something takes, the more you earn. That’s a terrible alignment of interests.
AI makes our position easier to justify, not harder. When we can deliver a project in a quarter of the time it would have taken three years ago, the value conversation becomes clearer: you’re paying for the result, the quality, and the twenty years of experience that ensures what gets built actually works. Not for how many hours someone sat in front of a screen.
The agency model is unbundling. The commodity work is separating from the strategic work. That’s uncomfortable if you built your business on commodity work dressed up as strategy. It’s liberating if you were always focused on the bit that matters.
We know which side we’re on. The question is: does your agency?
—
Versantus helps UK businesses navigate the AI transformation — honestly, strategically, and without the buzzword bingo. Let’s talk.