This week a major website builder did two things in the same breath. It cut about 1,000 people, roughly 20% of its staff, the biggest layoff in its history. And it told the world that the future of its business is a chat box where anyone can describe an app and get a working one back.
If you run an agency that builds on someone else's platform, stop and read that again.
The company didn't bury the reason in "efficiency" language either. It named AI out loud as a cause, which almost nobody in this industry does. That honesty is doing you a favor. It's showing you exactly what the platform values, and it isn't the assembly work you've been reselling.
The platform just told you what it values
The company's stock is down around 50% this year. Q1 revenue was up 14% to $541 million, but it posted a net loss of $57.5 million after a string of profitable quarters (CNBC, May 28). A strong shekel got some of the blame. AI got the rest.
Here's the part that matters for you. The same company spent $80 million last summer buying a six-month-old startup that turns plain-English prompts into deployed apps, with the database, the login system, and the hosting all handled for you. That little engine is now reportedly doing $150 million in annual recurring revenue, way ahead of its own internal projection (TechCrunch; the company's press room).
So while it laid off the people who do hands-on work, it poured money into the thing that lets your client skip you entirely.
Read that as a strategy, not a tragedy. The platform works for itself. It always has.
"We build on their platform" is not a moat
For years the easy agency play was simple. Pick a popular builder, learn its quirks, resell setup and templates to small businesses, charge a markup, maybe keep a hosting retainer.
That worked because assembly was annoying enough that people paid to skip it. The builder still had a learning curve. Connecting a form to a database still broke. Somebody had to know where the buttons were.
The prompt box just ate that gap. When the client can type "build me a booking site with payments and a customer login" and get something live, the part you were charging for is the part the platform now gives away to win the customer directly.
And it's not producing toys anymore. Apps built by non-developers on these engines have hit $1 million in annual revenue in three months (reported by The Infinite Loop / Nebius). The "it only makes junk" excuse is gone.
There's a line from one of those founders I keep coming back to. The moat was never the idea. It's the execution over time. Anyone can generate the artifact now. Almost nobody can own it, run it, secure it, and keep it alive for years.
Where the real work went
Here's the uncomfortable part, and I'd rather tell you straight. "Just move up to strategy" is not the safe answer either. The same AI that writes the code also writes the deck. Process theater and recycled strategic packaging are getting commoditized right alongside template assembly.
What doesn't get commoditized is the hard engineering and the relationship. Those are two different things and you need both.
The engineering: the prompt box ships code, but it doesn't review it. The security research piled up fast this year. Indirect prompt-injection attacks, an AI agent that wiped a production database despite a freeze order, attackers registering fake package names that AI tools hallucinate and then install. One development shop has even turned "clean up the broken AI-built app" into a paid service, where engineers audit the architecture, the security, and the dependencies after the fact. Their line: the code isn't the problem, the missing security review is. That's work the platform will never do for your client.
The relationship: judgment, accountability, somebody who picks up at 3am when the thing the client vibe-coded last Tuesday falls over in production. The platform is a supplier. It is not on the hook for the client's business. You can be.
What we actually do about it
We don't resell a builder. We own the engineering. When a client comes to us, what they get isn't a template, it's a system we can stand behind: hardened infrastructure, real security because we built it right the first time, and people who stay when something breaks.
For agencies, that's the whole point of how we work. Your brand, our engineering, we stay invisible. You keep the client relationship. We take the hard parts the platform treats as the first thing to automate away. We have white-label engagements running right now under NDA with a marketing agency, a PR firm, and a technology partner. It's not a slide in a pitch deck. It's in production.
Some plain advice if you're feeling this one in your gut.
Stop selling assembly. The floor under that price is gone and it's not coming back. Sell ownership, security, and the part where you're accountable for the outcome.
Move your best client contracts off hourly and onto retainer or outcome pricing before the commodity floor drops out from under your hours. When execution gets cheap, getting paid by the hour for execution is a losing trade.
And partner on the engineering you can't credibly own yet. That's not weakness. That's the difference between an agency that gets bought for a fraction of its revenue and one that's still standing in three years.
The platform showed you its hand. It will sell DIY straight to your clients and automate the assembly layer first. Fine. Build your business on the part it can't reach.
Talk to us
If your offer right now is "we build on someone else's platform" and this week made your stomach drop, that's worth a conversation. We do the engineering behind agencies all the time, invisibly, under your brand.
First conversation is free. No commitment. kief.studio/contact.