What Is Model Context Protocol (MCP) in Plain English?

The last time we overhauled our help docs, the featured images alone took a month because of all the manual labor involved. And that’s before approvals. Ask our social media manager Zachery, he lived through it.

We did it again a little while ago, but this time the workflow was waaaay different. I had Claude map every article, then I built the first three by hand so it could see my taste. The first pass was off brand, so I pointed it at our brand guide, which is already sitting in Atarim for every project. The second pass came back beautiful. Our logo in the corner, our backgrounds, real UI from the app.

Then I let it run the other hundred while I slept.

I woke up and they were done. On brand. Already uploaded to WordPress. Total attention from me: under an hour.

The thing that made that possible wasn’t a smarter model, but a Model Context Protocol — MCP — quietly doing the part I (and Zachery) used to do ourselves manually.

The copy-paste loop we all pretended was fine

Run that same job a year earlier and you’d have found me doing it by hand. Brand guide open in one tab, chat window in the other, typing out which logo, which background, which shade of purple. Once per image. A hundred times. Getting it slightly wrong on the ones I rushed.

That’s not an images problem, though. That’s every job.

Chat window open in one tab. The client’s site in another. You type a prompt, you wait, you copy the answer out, you paste it into WordPress or Figma or your PM board. And when the model needs to know which theme is active, which plugins are running, or how that custom post type is actually structured, you go and fetch that yourself and paste it in too.

You were the integration. You were the API. A human running between two rooms carrying context in a bucket, spilling half of it on the way. That missing loop is why AI didn’t actually speed up delivery for most of us in those early experiments.

So what is MCP, really?

The technical answer: MCP is an open standard that connects large language models directly to external data sources, tools, databases and content management systems, with a structured two-way protocol instead of a chat box. Mastering this requires understanding context engineering for web agencies so your tooling feeds models actionable environment state.

The version I’d give you over lunch: it’s the difference between describing your kitchen to a plumber over the phone and handing him the keys.

Same plumber. Same skills. Wildly different afternoon.

Under MCP, the AI stops being a clever stranger you brief and becomes something closer to a team member who can look for themselves. It can read your site’s parameters, see the active stylesheet, inspect the server config, and call tools you’ve explicitly approved — enabling role-based autonomous AI agents to work inside your client environments, not in a sandbox off to the side.

We’re not inventing anything here. I’ve always thought about this as digitizing human experiences. What MCP digitizes is the oldest move in agency life: walking over to someone’s desk, looking at their screen with them, and going “ah, right, now I see it.”

Why model context protocol for WordPress agencies changes the shape of the work

I’ve been dealing with creative website feedback for almost 18 years, and we’ve served over 200,000 websites through Atarim. Two things happen the moment context stops being something you carry by hand.

The AI stops guessing about the site. It doesn’t need you to explain how the page was built. It can query core status, list active plugins, read the database schema, look at the custom field setup on staging. It gets verified context before it proposes anything. Which means fewer confident suggestions about a plugin you don’t run and a theme hook that doesn’t exist.

Client feedback stops leaking on the way to the code. This is the one that matters to me most, because it’s the problem I’ve been chasing since the van.

Think about how feedback actually happens. Nobody writes a report. People point and say three to five words. “Make this bigger.” “Move this up.” “Change this color.” That’s human nature and there’s no point fighting it.

Normally those five words then travel: client to account manager, account manager to ticket, ticket to developer. By the time it lands, half the meaning is gone and someone’s guessing which button she meant. Capturing point-and-click client feedback directly eliminates this translation loss.

With MCP wired into a visual collaboration layer, the five words don’t travel alone. The screenshot, the coordinates, the page metadata, the element hierarchy, the brand guide, the project brief — all of it goes with them. Inside Atarim we end up layering something like 5,000 words of context onto a five-word request. That’s how a five-word request turns into the right CSS rule instead of a conversation.

Three MCP tools for web design and development you can put to work now

1. Populating and staging content without losing a week to it. Filling a WordPress build with structured content, custom post types and localised metadata is grunt work that eats developer hours and nobody’s proud of. An agent with MCP access can read your approved content deck, map fields to ACF or native blocks, and stage full draft pages with the schema in place. You review. You don’t type.

2. Debugging without the archaeology. An update breaks staging and suddenly a good developer is spending an afternoon reading error logs and deactivating plugins one by one. With an MCP server running in that environment, the agent pulls the debug log itself, finds the conflicting hook or the deprecated function, and stages a patch for a human to look at. Applying this across your portfolio enables automated fleet maintenance and regression testing without manual slog. You still make the call. You just skip the digging.

3. Design system to live CSS. Handoff is where design systems go to die. Connecting AI to WordPress workflows through MCP means an agent can read updated design tokens straight from your design source and push the matching CSS variables or theme config across several staging sites at once. The change happens once, in one language, everywhere.

Before you point this at a client’s live site

A couple of things I’d insist on, because “the AI can execute” and “the AI should execute unsupervised” are very different sentences.

Read freely, write carefully. Role-based access is the whole game. Let it read whatever it needs, autonomously, all day. Anything destructive or anything on a live site sits behind a human approval gate. Start read-only on staging with one project you’re not scared of. That’s your Monday morning move.

Pick open over walled. Every proprietary assistant wants your workflow to live inside its house. MCP is an open specification, which means the tools you connect today aren’t a bet on one vendor still being your favourite in two years. You keep the toolchain. That’s not a small thing.

Where this goes

The agencies that come out of this era ahead won’t be the ones writing the most prompts. They’ll be the ones whose CMS, design tools and client collaboration layer all speak the same language, so context stops being something a human carries around in their head and their clipboard.

That’s the shift from fiddling with AI to actually running on it.

We’re building Atarim’s MCP for exactly this: client feedback going in as a pin on a page, and coming out the other side as work that got done. If you want to try it, you can connect Atarim to Claude or Codex here and tell me what you break.

I’d love to know what you’ve hooked up so far. What’s the first thing you’d let it touch?

Ever finished the actual work on a site in an afternoon, then watched the project drag on for another two months?

That gap used to haunt me. For years I blamed myself for it.

Then an envelope turned up in the mail and it finally clicked.

Forty pages and a logo

We spent a full year chasing one client for the copy on a single About Us page. One page.

We did follow-up calls. Reminders. Everything either of us could think of.

Then one morning it showed up in an envelope. Forty pages of website copy, handwritten out, and on page one, the logo he wanted printed. 😅

Gosh, it was crazy. And the maddest part is that by then I wasn’t even shocked.

I’d built hundreds of sites out of a bedroom, then a spare room, then an office. The more we grew, the clearer one thing got.

The work itself was the fun part. Creating things was never what was killing us.

It was everything around the work. The waiting. The chasing. The endless back-and-forth.

The shelf

Here’s the shape of it, and once you see it you can’t unsee it.

Every time a task passes from one person to the next, it hits a shelf and sits there.

It’s like handing someone a baton and watching them put it down, because they’re mid-race on something else. Not out of laziness. They’re just busy with their own thing.

By the time they pick it up, a week’s gone. By the time it comes back to you, another week.

Now count the handoffs in a normal project. Brief to design. Design to client. Client feedback to you. You to dev. Dev back to client. Every arrow in that sequence is a shelf.

The actual work in there might be three days. The project takes two months.

And I know what you’re thinking, because I thought it too for years. That’s a client problem, right? Clients are slow. Clients don’t reply.

It isn’t. It’s just as true inside a team. Ask anyone who’s ever waited four days for a review from the person sitting eight feet away.

It’s even true when the team is just you. You put the thing down to handle something urgent, and the shelf is your own to-do list.

Where the money leaks out

The shelf is invisible, which is why it doesn’t show up in your numbers. But it costs you, and here’s exactly how.

The follow-up you meant to send never goes out. “I’ll do it tomorrow.” Tomorrow turns up with its own fires. That’s a proposal that never gets chased and a job you never find out you lost.

The marketing block you put in the diary for Tuesday gets skipped. There was client work sitting in front of you, and client work always wins. Which is exactly why the pipeline empties out three months later and nobody can work out why.

The monthly report you used to send quietly stops. We hope nobody notices, because clients don’t read them anyway, right?

That last one still gets me, because I did it too. And it was the stupidest thing I ever cut.

The report was never about being read. It was the one small thing that turned up before anyone asked for it. That’s most of the reason a retainer feels like value instead of a tax.

Calling it busywork is how you talk yourself out of the thing that’s keeping the client.

Why hiring didn’t fix it for me

The obvious answer to a capacity problem is more people. That’s what I did. And more people is genuinely better in a lot of ways.

But every person you add is another handoff. Another shelf. When my team grew, the number of things sitting and waiting grew with it, not down.

That’s the bit nobody warns you about. You hire to go faster, and you get a faster team with a longer queue.

Which is why I stopped thinking about this as a people problem and started thinking about it as a movement problem. The question isn’t “who does the work.” It’s “what happens to the work in between the people.”

Everything we’ve built since comes from sitting with that question. Not how do you build a website faster, because building was never the slow part. How do you stop the work from sitting down.

Something that picks the baton up the moment it’s put down, at 4pm on a Friday, on the night you’re at your kid’s thing, is a different kind of team member than another pair of hands. It doesn’t have a queue of its own.

So if your projects drag even when the work is easy

Hear me out on this one, because I wish someone had said it to me a decade earlier.

It isn’t you. It never was.

You are not slow. You are not disorganised. You are running a relay where half the runners are standing still, and you’ve been blaming your own legs.

It was never the work. It was everything around it.

Next up I’ll show you what that gap was actually costing all of us, in money. I sat down with more than fifty agency owners to work it out, and the number was worse than I expected.

Where’s the shelf in your projects? I’d bet you already know which handoff it is. 🤘

When I started my agency, there was a sentence that paid for a lot of months.

“Your site isn’t mobile friendly.”

That was £700, right there, and it was true for basically every site we’d ever touched. Not a hard sell. Not an invented problem. A real thing that mattered, that the client couldn’t see, that we could fix.

Two years later it was a different sentence: “your site isn’t GDPR compliant.” £700.

Then Google changed the performance rules. £700.

Then accessibility, and now you’re breaking the law if you ignore it. £700.

Same shape every time. The industry hands you one thing that’s true of every client on your book, and for about eighteen months there’s a queue of work sitting there that nobody has to be convinced of.

Right now it’s agents

None of your clients’ sites are ready for them.

Not the content structure, not the way the information is organised, not what a machine reading the page can actually understand and act on. It’s the same situation as 2013 and a phone screen: real, technical, invisible to the client, and about to be obvious to everyone.

Which of your clients need that work? All of them.

How many have you told? 😅

That’s not a dig. It’s the actual pattern, and I did it too. By the time everyone’s talking about it, somebody else has already made the call.

The bit that costs more than the big ones

The industry-wide job is the easy example, because it’s loud enough that you eventually notice it.

The expensive one is the small stuff. The things you’d spot in ten minutes if you had ten minutes.

A booking form for the yoga studio client, because they’re still doing it over DMs. Better SEO for the restaurant, because they’ve dropped off page one and haven’t mentioned it. A site that’s got slower over three years, in the way sites do, a plugin at a time.

Every single one of those is work the client wants. You already know it’s needed — you’d say so instantly if someone put the site in front of you. And it’s much easier to sell to a client who already trusts you than to go and find a new one.

So why doesn’t it happen?

Because nobody looks. Not out of laziness — because looking is unpaid, unscheduled work, and the reactive queue is neither. There’s always something in front of it with a name and a deadline attached.

The opportunities have always been there. You’ve just been blocked from capturing them, and from charging for them.

What “getting ahead” actually requires

Here’s the honest version of what this takes, if you do it by hand.

For each client: open the site. Check performance properly, not a vibe. Look at what’s ranking and what isn’t. Look at what they’re doing manually that could be a form. Check what’s out of date. Compare all of that against what this particular client is actually trying to achieve, because a booking form matters enormously to the yoga studio and not at all to the accountant.

Then write it up. Scope it. Price it at your rate. Make it something the client can say yes to without a meeting.

Call it forty minutes a client, if you’re quick and you know the site well.

Ten clients is most of a day. Every week, so it’s actually current rather than a snapshot from March. Nobody does this. I never did it, and I knew exactly how much it was worth.

That’s the whole reason the reactive model wins. Not because anyone chose it.

The version that fits in a Tuesday

This is the part that changed for me, and it’s why Claro exists as a project manager rather than a chat box.

You can point him at a client project and ask him to go and look — against that client’s success criteria, not a generic checklist — and come back with what he found, scoped and priced at your rate.

If you want to try it on a real client, drag the project into the chat and paste this:

Hey Claro, apply the Upsell Opportunity workflow to the client I just dragged into this chat. Create a plan, scope it up and quote it based on my hourly rate.

Then do the thing I’d ask of anyone: check it properly. Some of what comes back you’ll bin, because you know something about that client that isn’t written down anywhere. That’s fine, and it’s the job. You’re reviewing a scoped proposal rather than generating one from a blank page, and those are completely different Tuesdays.

Now do the maths on running that across your whole book every week, and how long that would take you by hand.

Why this matters more than the money

The upsell framing undersells what’s actually happening here.

When you turn up with something the client hadn’t thought of, the relationship changes. You stop being the person who does the tasks they send you and become the person looking after their site. Those are different businesses with different prices, and only one of them survives a budget review.

The client can’t tell the difference between a good agency and a cheap one from the work itself — they mostly can’t read the work. They can absolutely tell the difference between someone who noticed and someone who waited.

That’s what the £700 jobs were always really about. Not the mobile fix. The fact that you were the one who mentioned it.

Pick your worst-performing client site — the one you’ve been quietly avoiding — and go and look at it properly this week.

I’d bet there’s a job on it you’d have found in ten minutes. What is it? 🤘

Who made Jaws?

Steven Spielberg, right?

Except he didn’t hold the camera. He didn’t build the sets. He didn’t cut the film. Hundreds of people did all of that, and most of us couldn’t name one of them.

The taste, the final call, the name on the poster? His.

And nobody has ever once said the director didn’t make the movie.

I think about that a lot, because there’s a version of this argument that agency owners have with themselves constantly and almost always lose.

The day I stopped doing the work

I learned this years before AI turned up, on the day I made my first hire.

He was a designer, and he was good. And the first thing that happened was that I felt slightly fraudulent. Work was going out under my agency’s name that I hadn’t drawn. I’d briefed it, I’d set the standard, I’d sent it back twice before I was happy with it. But my hands weren’t on it.

It took me an embarrassingly long time to understand what had actually happened that day.

That was the day I stopped getting paid for my hands and started getting paid for my judgement.

Which is a promotion. It just doesn’t feel like one at the time, because the thing you were proud of — being the one who could do it — quietly stops being the thing you’re valuable for.

That’s where a lot of people are sitting right now with AI, and it’s why the conversation gets so emotional. It isn’t really a conversation about tools. It’s a conversation about what you are if you’re not the one doing it.

Here’s my answer, and I’d stand on it: you’re the director. The work gets done, you set the standard, and it ships under your name only when you’re happy with it. That was true when the hands belonged to a designer on payroll, and it’s true now.

What the hire also brought

Now, the honest part, because I don’t want to make hiring sound like it solved everything.

That first hire didn’t just bring a payroll. He brought a queue.

Every piece of work now had a moment where it sat, waiting on one of us to pick it back up. He’d finish something and it would wait for my review. I’d send notes and they’d wait for his next free morning. Neither of us was slacking. We were both just mid-something-else, permanently.

And when more people joined, there was more waiting, not less. Every new person is another place work stops.

So the trade was real but it wasn’t free. I bought hands and I paid in delay. Everyone who’s ever grown a small team knows this feeling and mostly assumes it’s a management failure. It isn’t. It’s arithmetic.

What I wanted — what I think most of us wanted — was the hands without the queue.

Hands without the queue

That’s the thing I’ve been building for the last few years, and it’s why Atarim isn’t shaped like a tool with an AI feature bolted on. It’s shaped like a team you direct.

Practically, that means it has to actually know the environment. It has the deepest connection to WordPress on the market, it supports the top 120 plugins, and we’ve written hundreds of skills that teach it to use them the way an agency would — not the way a general-purpose model guesses at them.

That’s the difference between something that can describe what should be done and something that goes and does it. And it’s why people tell me it’s the first plugin they install on a new project, because it’s the thing they need in order to do everything else.

You can also run it through Claude, Codex, or whatever you already use, by installing the Atarim MCP. Same agency, your AI, your workflow, your credits. I’d rather meet you where you work than ask you to move.

What changes when you direct instead of do

Three things, and I’ll be careful here because this is where marketing usually starts making promises it can’t keep.

You stop trading hours. Not automatically — you have to actually change how you quote. But when the labour isn’t yours, pricing the outcome stops being an aspiration you read about in a book and becomes the obvious thing to do. That effective hourly rate I keep going on about — the one that lands around $7.50 to $13 for most agency owners once you count every unpaid hour — that number is a fraction, and this changes the bottom of it.

Your clients trust you more, not less. This surprises people. The fear is that clients will feel fobbed off. What actually happens is that something turns up before they ask, every week, whether or not you were at your desk. That’s the behaviour of an agency that’s paying attention. They don’t experience the mechanism, they experience being looked after.

You get to choose your size. Right now that isn’t really a choice, is it? It’s whatever your hours allow. Growing means hiring, which means management, which means you do less of the thing you’re good at. Staying solo means staying small. Take that ceiling off and both of those stop being true, and you get to pick based on what you actually want your life to look like rather than what the arithmetic permits.

I’m not going to tell you which one to pick. I’ve done both and there’s no correct answer.

The Monday test

Here’s how I’d know if any of this were real, and it’s the test I’d apply to anything claiming to change how you work.

Picture a Monday. You open the laptop and the first draft of the week’s work is already done, waiting on your review.

Same clients. Same standard. Same name on the work.

How does that feel?

If it feels like a relief, you’re a director who’s been stuck doing crew work, and you’ve probably known that for a while.

If it feels like a loss, that’s worth sitting with too, and I don’t mean that dismissively. I felt it on my first hire. The thing you’re mourning is real. It’s just not the thing anyone’s paying you for.

Spielberg didn’t hold the camera. 🤘

Let me be straight about something I’m hearing a lot lately.

I had a call with a member of our community recently, someone who’s been doing this a long time. She’d lost a job to the client’s nephew, who’d built a site in an afternoon.

Not a competitor. Not a cheaper agency. The nephew.

And meanwhile there’s a new influencer every week promising a full website in minutes, with a screen recording to prove it.

So your quote lands, and it doesn’t just look expensive. It looks like a robbery.

Why the usual defences don’t work any more

The instinct is to explain. To educate the client on what actually goes into a site — the strategy, the structure, the testing, the things they can’t see. I’ve given that talk many times and I still believe in it.

But it’s getting harder, and I want to be honest about why.

It used to work because the gap between what the client imagined and what the work required was enormous, and closing that gap was genuinely persuasive. “It looks like five pages, it’s actually forty hours” landed, because they had no counter-evidence.

Now they have counter-evidence. They watched their nephew do it. They saw the video. The gap has genuinely narrowed for the thing they’re picturing — which is a site that exists and looks fine.

You can argue that the nephew’s site is bad. You’d probably be right. But you’re now arguing about quality with someone who can’t assess quality, holding a quote that’s twenty times the price of a thing that visibly exists.

That’s not a winnable conversation. It’s the wrong conversation.

Selling hours is what’s actually broken

Here’s the thing underneath all of it.

The old model prices your time. Whatever the invoice says, the logic is hours — that’s why the client’s mental comparison is “how long would this take,” and that’s exactly the comparison the nephew wins.

If your value proposition is that you do the work, you are now competing on how long the work takes. And the work is getting faster for everyone, including the nephew.

For most agencies that’s the thing that slowly kills them. Not one dramatic lost pitch. A gradual squeeze where every quote needs more justification than the last.

But the old model only has to matter to you if you’re still selling hours.

What you can say instead

There’s a sentence I think is worth building a business around, and it takes some getting used to:

“I don’t do the work. I own the outcome.”

Read that again, because the first reaction is usually that it sounds like an admission. It isn’t. It’s the same thing every serious business in every other industry sells.

You don’t hire a builder because they personally lay every brick. You hire them because if the wall falls down, it’s their problem. You’re not buying their hands. You’re buying the fact that they carry the risk, and that they’ll notice things you wouldn’t.

Nadine, one of the people I’ve been speaking to about this, put it better than I have: high-value clients aren’t buying software, they’re buying accountability.

That’s what the nephew can’t sell. Not because he’s not talented — he might be. Because when the contact form breaks in November, he’s back at university, and nobody is accountable for the thing he made.

When you sell the outcome, the price conversation stops being the same conversation. You’re not defending a number against a faster alternative. You’re the only person in the conversation offering to be responsible.

The bit that makes it true

Now, I’m not going to tell you to just say that sentence, because saying it isn’t the hard part. Meaning it is.

If you say “I own the outcome” and then only respond to what lands in your inbox, the client will work it out within a quarter. Owning the outcome means noticing things before they do. It means the site not silently getting slower. It means the thing that turns up in March that they hadn’t thought about.

And that has always been the catch, hasn’t it? That’s a full-time job per client, and you have twelve.

That’s the actual reason most agencies sell hours. Not because they prefer it. Because owning the outcome across a whole book was impossible for a small team, so hours were what was left.

It’s the specific thing I’ve spent the last few years trying to fix. Your AI agency doesn’t just remember your five or six best clients, the way you do — it knows every project on your book, and you can ask it anything. It finds work without being asked, drafts and adjusts pages, and has things ready for your review while you sleep. Not so you can do more hours. So that “I own the outcome” is a description of what’s happening rather than a line in a pitch.

You still make every call. Nothing goes near a client without you seeing it. That part isn’t negotiable and never will be — it’s the entire thing you’re selling.

One thing to do this week

Don’t try to learn a whole new business model this week. I mean that.

Put one real client through it. Open their project, pick the workflows that make sense for them, set the cadence. Then ask Claro where else you could be adding value on that account.

One client. See what comes back. Argue with it.

Then, the next time a quote goes out, try the sentence. Not as a script — just notice whether the conversation goes differently when you’re selling accountability instead of hours.

Tell me how it goes, good or bad. Genuinely, the bad ones are more useful, and what people send back is what shapes what we build next.

Now go be it. 🤘

Before Atarim, I ran a web agency in London. Before that I toured Europe in a band and built client sites from the back of a van, over stolen McDonald’s WiFi, way before 4G was a thing.

I turned the van into an agency. We built hundreds of websites and handled every 2am launch you can imagine.

So I have a fair idea of what your week looks like.

And I know something about the way we all describe ourselves, because I did it too. The agency in my proposal was always a bit bigger than the agency that actually existed.

The proposal said design, development, SEO, performance, accessibility, ongoing care. The reality was me, a few people, and a list of things we kept meaning to get properly good at.

The line items I paid for instead of hiring for

There were parts of the job I genuinely couldn’t do at the level a client deserved. So I bought them in.

I used to pay a specialist $800 for a proper speed pass. A couple of grand for the SEO groundwork on a decent build.

That was the honest version of “we do performance and SEO.” I outsourced it, marked it up a bit, and moved on.

And that was only the stuff clients could see. Underneath there was a whole second list. The pre-launch QA that never got done as thoroughly as it should have. The security scan I meant to schedule. The content that went stale on sites we’d built two years earlier and nobody had looked at since.

The monthly report is the one that still gets me. We used to send them. Then we stopped, because who reads them anyway, right?

That was the stupidest thing I ever cut. The report was never about being read. It was the one small thing that turned up before anyone asked for it. That’s most of the reason a retainer feels like value instead of a tax.

Calling it busywork is how you talk yourself out of the thing that keeps the client.

Everybody’s list is the same shape

I’ve spoken to a lot of agency owners since 2019, when I built WP Feedback to fix my own agency and thousands of other agencies ended up running on it. Which meant I suddenly had a lot of people I could just ask.

The list is remarkably consistent. Everyone has three or four things they’re genuinely great at, and then a much longer list of things they know matter, know they’re not doing well, and can’t justify hiring for.

Because the maths never works. A proper SEO person is a salary. A dedicated QA person is a salary. Someone whose whole job is watching your existing clients for things that need attention? Nobody hires that person. Not at our size.

So the list stays a list. And every year it gets a bit longer, because the industry keeps adding to it. Mobile. GDPR. Core Web Vitals. Accessibility. Now agents.

None of this is a skills problem. You could learn any of it. I learned a lot of it the slow, expensive way. It’s a capacity problem, and capacity problems don’t get solved by trying harder on a Sunday night.

What actually changed

Here’s the bit I find genuinely exciting, and I’ll be careful not to oversell it.

Every one of those jobs is something the best agencies were already doing by hand. The pre-launch QA checklist. The quarterly content refresh. The monthly report. The security scan. The upsell conversation that starts with “I noticed something on your site.”

None of it is new. None of it is clever. It’s the boring, repeatable work that separates an agency that keeps clients from one that keeps chasing them.

We’ve spent six years watching how our users actually do this work, and the whole of our AI implementation comes out of that. I’ve always thought about it as digitizing human experiences. We’re not inventing a new way to run an agency. We took the way good agencies already run and taught a machine to do it at a scale a small team could never staff for.

That’s what an AI agency is. Not a robot that writes copy. A set of the jobs you already know you should be doing, running without you having to be the one holding the broom.

It learns your services, your rates, your brand. It gets to know your clients the way you do. You give it work, or you let it bring you work you didn’t ask for.

The parts of the craft you love stay yours. Nothing ships without your say so.

Something to try this week

If you’ve got Atarim open, drag one client project into the chat and paste this in:

Suggest the best workflows (SOPs) for MY agency. I want a repeatable plan I can onboard new and existing clients to that will make me look like a rockstar.

Or, if you want the faster hit of seeing what it actually knows about your client:

Look at this client’s logo and tell me honestly whether it’s dated. If it is, show me what I should pitch them instead.

Only got one client in there so far? Start with that one. Then add a few more, because the interesting stuff starts when it can look across your whole book and tell you which client needs you before they email you.

That’s the part I’d have paid real money for in my agency days. Not another tool that helps me build faster. Something that noticed things while I was busy.

I’m curious about your list, actually. What’s the job you’ve been meaning to hire for and never have?

Hit reply and tell me. I read every one, and I answer them. 💜

I’ve been building websites since I was 14, back on GeoCities (remember that?), so I’ve watched this industry panic before. Quite a few times, actually.

Drag-and-drop builders were going to kill us.

Then WordPress was going to kill us.

Then the page builders.

Then Figma. 

And every single time, the opposite has happened. The only ones who lost out were the ones who misread the shift.

I’ll admit, this one does feel much bigger. Your clients are getting access to tools like Claude, Lovable, and Base44, and they can make websites themselves instantly.

So yeah, the worry is real. A survey of 500 US web designers this year (nice work by the team at 20i) put numbers on it:

• 76% named AI as their single biggest worry about the future of the work.
• 75% said the competitive pressure is already showing up in their business.

But read the rest of that same survey, because the panic is only telling you half of it. The exact same designers who are worried also still feel paid:

• 78.6% feel properly paid for their work.
• 37% are clearing over six figures a year.
• 9 out of 10 say their clients underestimate what web work actually costs.
• Agencies leaning into AI are growing revenue by 58%.

That’s not a profession in collapse. It’s a profession splitting in two: the work a machine can do on its own, and the work it can’t. And the DIY corner of the web, the part everyone’s terrified of, is tiny and always has been.

Around 95% of the world’s websites are still built and run as a professional service.

It took me a while to come around on this too. Then I saw where the shift actually was.

Everyone’s staring at these instant websites like that’s the battlefield. But it’s the wrong place to look.

Building was never the hard part. What took me years wasn’t dragging a hero section into place. It was learning to run the business: the SOPs, the pricing, the client management, the stuff that separates a freelancer from an actual business.

I used to pay a specialist $800 for a proper speed pass, the performance and Core Web Vitals clean-up, and a couple grand for the SEO groundwork. I learned it all the slow expensive way. I’d bet your list looks a lot like mine.

All that hard-won knowledge? It’s a workflow now. It runs from day one for someone who walked into this industry last month.

From four years down to an hour of setup. That’s the shift. And it means the workflows themselves can’t be your edge anymore, because everyone starts with them now.

Which brings me to the uncomfortable bit, and the reason some agencies make it through this and some don’t.

AI is making two kinds of agency right now: the reactive agency and the proactive agency.

To the reactive agency, the monthly retainer feels like insurance. “I’m here if it breaks. I’ve got the tooling set up. I jump in when it matters.” To the client, that exact same retainer feels like a tax. “I’m paying a hundred bucks a month for a guy I haven’t heard from since March and I don’t know why.”

And when an AI tool shows up offering to do it now, for less, that client was already half out the door before the AI ever showed up.

The reactive agency is the yes-man. Waits to be asked, then scrambles, then bills for the time.

The proactive agency moves first. Sends the report before anyone asks for it. Flags the broken contact form before the client notices they’re bleeding mobile leads. Bills for outcomes, not hours.

Being proactive used to be a luxury. Updating 40 client sites every Friday ate most of a day, so even the best of us saved it for the top few clients. Now it costs almost nothing, so you can do it for everyone.

Which is the whole reason we built V5 the way we did. 

It’s the infrastructure underneath the proactive agency. Around 200 workflows that run from the moment you switch it on, with your name on them. The Monday report that goes out before anyone asks for it. That same $800 speed pass, running on every client site instead of the two who could afford it. The same SOPs that took me four years and a lot of money to learn.

If you haven’t tried Atarim V5 yet, this is your chance to be the proactive agency.

Let the machine have the commodity work. You move up a level, to the taste and the judgment and the calls it still can’t make on its own.

Would love to know what you’re seeing in the design space. Am I reading any of this wrong?