Every Tech Wave Made Agencies Bigger. AI Will Too.

There's an irony in agencies being the anxious ones about AI. For forty years, agencies have been the first to benefit from every technology wave. Not survive them. Grow because of them.

The pattern is so consistent it's almost boring, and it's worth walking through, because the "AI kills agencies" take makes the same mistake every previous take made.

Wave 1: desktop publishing, mid-1980s

Before the Mac and PageMaker in 1985, a brochure meant a typesetter, a paste-up artist, and a trip to the print shop. Weeks, and thousands of dollars. Small businesses made do with typewritten flyers.

Desktop publishing collapsed the production cost. The prediction at the time was that typesetters and layout artists would be out of work and agencies would shrink because clients could do it themselves.

What happened instead: every small business suddenly wanted a logo, a brochure, a newsletter, because they could finally afford one. And most of them discovered within a week that owning PageMaker didn't make them a designer. Demand for design work exploded through the late eighties and nineties. The technology removed some production roles and created far more design work than it destroyed.

Wave 2: the internet, mid-1990s

The early web was supposed to disintermediate everything. Build your own site (GeoCities, anyone?), run your own email campaigns, sell direct. Agencies would be irrelevant because the tools were so easy.

The reality is that the internet created the modern agency industry. Before 1995 there was no such category as a digital agency. By 2000 there were thousands.

The numbers here are public. The IAB's internet advertising revenue reports tracked US digital ad spend at $267 million in 1996, $12.5 billion in 2005, $59.6 billion in 2015, and well past $200 billion by 2024. Every one of those dollars was work for someone, and agencies took a large share of it.

And the "anyone can build a website" argument has now been running for thirty years. Squarespace, Wix, WordPress. All good tools. All supposedly the end of web agencies. Web agencies kept multiplying, because the bar for "good enough" keeps rising and businesses keep needing help to clear it.

Wave 3: social media, late 2000s

Same script. Facebook, Twitter, Instagram, free to use. Every brand its own publisher. Why would you need an agency?

Because running social media well is a full-time job, and most companies found that out about three months in. Content calendars, community management, paid promotion, analytics, crisis response, algorithm changes, the endless appetite for new material. Social media management didn't exist as a service line in 2007. Within a decade it was a standard department at every agency and a category of its own. The pitch went from "you should be on Facebook" to "we'll run your whole social presence" in about eighteen months, because clients went from curious to overwhelmed that fast.

Wave 4: mobile, early 2010s

iPhone in 2007, App Store in 2008, and suddenly everyone needed a mobile strategy. Then a responsive site. Then an app. Then a mobile-first redesign. Then a progressive web app.

Each sub-wave brought its own surge of work. The tools got easier and the expectations got higher at the same time. By the time a decent responsive site was a template away, clients wanted custom mobile experiences with animation, offline support, and sub-second loads. The Bureau of Labor Statistics has web developer employment growing strongly through the 2010s and still projecting above-average growth. Mobile didn't eat those jobs. It added to them.

Wave 5: marketing automation and data, mid-2010s

HubSpot, Marketo, Salesforce Marketing Cloud, Google Analytics, the Facebook pixel. Tools that would automate marketing so thoroughly that outside help would be unnecessary.

Instead they created a consulting industry. Marketing operations. CRM implementation. Attribution. Tag management. Companies bought $50,000-a-year platforms and then discovered they needed someone to run them, and that someone was usually an agency. Plenty of agencies built a third of their revenue on HubSpot implementations inside two years of that being a thing at all. The tool created the work.

The four stages, every time

The tool appears. Everyone experiments. "Anyone can do this now." Panic about obsolescence.

Cost collapses. What cost $50,000 costs $500. What took 600 hours takes 60.

Demand explodes. Because it's cheaper and faster, far more people want it, including markets that couldn't afford professional work before.

Complexity moves up. The easy version becomes table stakes. Competitive advantage shifts to the harder, more strategic, more human layer on top. That's where agencies live.

AI is in stage two, heading into three. The cost of a first draft, a set of design concepts, a data pull, a basic automation: all dropping fast. What comes next is more demand, from more clients, for more work, at a higher level.

The agencies that will struggle are the ones selling stage-two work at stage-one prices. If the whole value proposition is "we write blog posts," then yes, AI is a threat. If it's "we build content programs that produce qualified leads, and we use every tool available to execute them better than your in-house team," you're fine. Better than fine.

The mistake every anxious take makes

It assumes a fixed amount of work in the world. A set pile of tasks, and AI takes from the pile, so people get less.

The pile grows. That's the whole history above.

When spreadsheets automated accounting arithmetic in the eighties, accountant employment went up, not down, because analysis became affordable for smaller businesses, new kinds of analysis became practical, and the standard for good financial management rose. More work, more jobs.

The same thing is happening now. The agency that produced one video a month for a client can produce four. The agency that could only serve $50,000 clients profitably can now serve $15,000 ones. The agency that spent twenty hours on reporting can spend them on strategy.

More output, more clients, more value per hour. That's not a threat. That's the growth engine, again.

If you want the practical version of this argument, how agencies are actually using AI assistants is the companion piece.