Articles Article

Head in the Sand or Hair on Fire

Most agencies I talk to are standing somewhere on a line. At one end, nothing changes. At the other end, everything changes at once, with nobody holding the wheel.

4 min read

Head in the Sand or Hair on Fire

Both ends feel safe to the people standing on them. Both are expensive. And the bill for each arrives later than you think, which is why so many good agencies are drifting toward one or the other without noticing.

This piece is about those two ends, and the ground in the middle where I think the smart money sits.

At one end: head in the sand

The first position sounds sensible. There is plenty of WordPress-style work out there. Clients keep paying for plugin updates, theme patches and the odd security scare. Why rock the boat?

So the agency keeps triaging. A plugin breaks, someone fixes it. A client wants a new landing page, someone bolts it on. The site gets heavier, slower and the margin gets thinner, but the invoices still go out, so everything feels kinda fine as long as you’re not on Reddit.

Then comes the classic move. The agency lead asks the dev team whether they want to change how they work. That is like asking turkeys if they fancy an early Christmas. Of course the answer is no.

I don't say that to knock developers. They are rational. The current stack is what they know, it is where their value sits, and change looks like risk to them personally. But the question was never theirs alone to answer. It is a business decision about margin, speed and where the agency wants to be in three years.

Here is what the head-in-the-sand position costs:

  • Margin leaks into maintenance work clients resent paying for

  • Front-end and design talent spend their days on plumbing, not craft

  • Every site is a one-off, so nothing compounds

  • Competitors using AI well quote faster and lower, and clients notice

  • The longer you wait, the bigger the migration when you finally move

Standing still is a choice. It only looks like the absence of one.

At the other end: hair on fire

The other end looks like the opposite. Agencies hurling themselves into AI, prompting sites into existence, shipping in hours what used to take weeks. It feels brave. It feels like the future.

And in fairness, the speed is real. AI is amazing. But speed without structure is how you build a future shit show, one project at a time.

Ask a few simple questions of a fully AI-led agency six months in:

  • Where does each project live? Which tool, which account, which hosting?

  • Who authored it? A person, a model, a freelancer's personal subscription?

  • Which model and which prompts produced it, and can anyone reproduce that?

  • When something breaks at 4pm on a Friday, can a human open the code and fix it?

  • What happens when token prices jump, or a vendor changes its terms?

If the answers are "not sure", that agency has traded one kind of mess for a faster, harder-to-see kind. I call it AI spaghetti. Every project works, until it doesn't, and then nobody knows which strand to pull.

Clients won't care that it was fast. They will care that it broke, that nobody can explain why, and that the fix costs more than the build.

What both ends have in common

At first glance these two realities look like opposites. They’re not. Both ends hand control away from the agency.

The head-in-the-sand agency hands control to a legacy stack and a plugin ecosystem it doesn't own. The hair-on-fire agency hands control to whichever AI tool is hot this month, and to whoever happened to prompt it.

Either way, the chaos and the cost live in the gaps: between tools, between people, between what was built and what anyone can maintain. We went deep on this in our BluPaper, Compounding AI Chaos.

The sweet spot: the Agency Web Ecosystem

The middle of the line is not a compromise. It is the one position where the agency keeps control of its creative output, every project and every contributor, and still gets production speed from the AI model that suits it best.

That is why we built Blutui as an Agency Web Ecosystem, inside a working agency, long before AI was the headline. The idea is simple. Agentic where it saves time, human where it counts, and the agency in charge of both.

In practice here’s what that looks like:

  • Every project lives in one place, on managed hosting under the agency's own brand, with seamless access to new features, releases and updates

  • Real code, owned by the agency, that a human developer can open and fix

  • AI connected through MCP to the models you choose, not a model you're locked into

  • Predictable costs, with code access as a hedge if token prices climb

  • Clients edit content through Canopy Blocks without touching anything they could break

  • Front-end and design talent spend their time on craft, while repetitive production runs in the background

That last point is the one I care about most. AI should make agency talent go further, not replace the judgement that makes an agency worth hiring.

So, where do you stand on this?

If you lead an agency, try three questions on for size this week:

  • If we keep doing exactly what we do now, what does our margin look like in two years? Or, more likely, six months?

  • If half our sites were built by AI tomorrow, could we say where each one lives and who can fix it?

  • Who in the agency is making this call: the business, or the people most comfortable with the status quo?

There is no prize for standing still, and no prize for running fastest into a mess. The agencies that win the next few years will be the ones that move with intent, with their hands firmly on the steering wheel.

If you want to see what the sweet spot looks like in practice, book a Fly-Through here. Happy to show you around.