The Agency That Stops Charging for Content Wins

A few weeks ago, Sequoia Capital partner Julien Bek dropped an essay that made the rounds everywhere. His thesis: the next trillion-dollar company won't sell software. It will sell the work itself. A software company masquerading as a services firm.

For every dollar businesses spend on software, they spend six on services. AI is coming for that budget.

He's talking about insurance brokerage, accounting firms, legal contract work. Makes sense. But I read it and kept thinking about a different industry entirely.

Agencies.

The uncomfortable math

I was in a meeting last month with an agency owner. Good agency, strong brand clients, solid reputation. He walked me through a campaign they'd just delivered. Beautiful work. The kind of thing you'd put in a case study.

Then I asked how many hours went into it.

Most of the budget, and I mean the vast majority, wasn't the big idea. It wasn't the strategy. It wasn't even the design. It was adaptation. Resizing. Localising. Reformatting for sixteen different markets. Re-exporting because legal changed a disclaimer in Belgium.

He knew it too. He just shrugged and said: "That's the business."

And he's right. That IS the business. The question is whether it should be.

Bek's framework, applied sideways

Bek draws a line between copilots and autopilots. A copilot makes a professional faster. Think AI inside Photoshop. The designer still does the work. An autopilot sells the work itself. No professional needed for the production part. You ask for the output, you get the output.

His key insight: when a task is already outsourced, AI doesn't need to change behaviour. It just needs to change the supplier.

Agencies are outsourced by definition. Brands already accepted that this work can be done externally. There's a budget line for it. They're buying outcomes. A campaign, a set of assets, a seasonal rollout.

So what happens when the supplier changes?

The thing agencies keep doing wrong

I've had this conversation maybe forty times this year. An agency gets excited about automation. They buy into the vision. They start exploring.

And then they do the safe thing.

They keep designing in their existing tools. Then they template some of it. Then they generate some variants. Then someone manually QAs everything. Then it goes into a DAM. Then someone exports for each channel.

Same workflow. One step got faster.

They're protecting their billing model. If content production becomes fast and cheap, they lose the hours they're selling. So they keep automation in a box. Controlled. Harmless.

I get it. I really do. But they're optimising for a business model that's about to get disrupted by someone who doesn't care about protecting it.

What the radical ones are doing instead

There are a handful of agencies, small number but growing, that looked at all this and drew a very different conclusion.

They stopped charging for content production.

Not reduced. Stopped.

They do the design work. They build the strategy, the brand system, the campaign architecture. Every rule, every ratio, every safe zone. That's where the value lives and they charge accordingly.

Then they make all of it available through technology, through our platform CHILI GraFx, so brands can produce content at scale. Every format, every language, every market. Unlimited.

The production is free.

Because they own distribution.

Print. Digital. In-store. Social. Signage. POS. They control where the content goes, how it gets there, and they make their money on that. Both print and digital.

When I first saw this happening, I'll admit I was surprised. Not by the technology. We built the technology. But by the business model courage. These agencies essentially looked at their own P&L and said: the thing generating most of our revenue right now is the thing we need to give away.

That takes guts.

Why distribution is where the money actually lives

Most people underestimate what it takes to get a campaign into the world.

Designing it is hard. Producing it at scale is harder. But getting it into 40 countries, across 200 touchpoints, correctly, with the right pre-flight specs, colour profiles, substrate requirements, bleed settings, responsive sizing, platform rules, metadata, accessibility standards... that's where everything falls apart.

Brands can't do this themselves. Never could. That's why they pay agencies and production houses and print brokers and media buyers.

The agency that automates not just the content but the entire output chain, and controls the distribution infrastructure, becomes very hard to replace.

You can find another designer. You can't easily replace the system that delivers your entire seasonal campaign overnight.

Now add an LLM to this

This is the part I keep thinking about.

What if an agency doesn't just template its creative work, but makes it queryable? The brand system, the campaign logic, the design rules. All of it structured so that an AI agent can work with it.

A brand manager in São Paulo types: "I need the 'Share a Drink' summer campaign for Brazil. Portuguese. Local pricing. All social formats plus the in-store lightbox."

And it just... happens.

No email. No brief. No three-week wait. No change order.

The agent knows the campaign architecture. It knows the brand rules. It generates through the automated design system. It routes to print and digital.

No extra cost for the production. The agency already made its money on the design, the strategy, and the distribution infrastructure.

The technology for this exists. CHILI GraFx already does the structured, rule-based, automated content production. The LLM layer, the agent interface, is arriving now. And when it connects to a proper design system, it doesn't hallucinate. It produces on-brand, production-ready content. Because the rules were already defined by humans who know what they're doing.

So what's the actual business model?

The cleanest version I've seen goes something like this:

The agency charges a retainer for creative strategy and design systems. That's the intellectual property, the thinking, the brand architecture. That's the expensive part and it should be. Then content production is unlimited and automated. Essentially a utility. And distribution, the logistics of getting the right content to the right channel in the right format, is where the ongoing revenue comes from. Per market, per channel, per campaign cycle.

Some agencies are going even further. Leading with distribution. Treating the creative almost as a loss leader because the real margin is in placement and fulfilment. Others are moving toward pure outcome pricing. "Run our summer campaign across Europe, fixed price, we handle everything."

Each version is different but they share one assumption: content production is no longer a profit centre. It's electricity. You don't sell it. You use it to power everything else.

Most agencies won't do this

I know. I've been in this industry long enough.

They won't do it because their P&L is built on production hours. Because their teams are structured around manual workflows. Because their contracts are time-and-materials.

And honestly, because it's terrifying to give away the thing you've been selling for twenty years.

But Bek's essay should keep agency owners up at night. The message from arguably the most influential venture firm in the world: companies that sell work will scale. Companies that sell tools, or hours, won't.

The agency model as we know it sells hours.

The ones that figure out how to sell outcomes instead, design, strategy, distribution, won't just survive. They'll scale in ways the old model physically can't. Because in the old model you scale by hiring. In the new one you scale by deploying.

One last thought

The next trillion-dollar company probably won't be an agency. Let's not get carried away.

But the next agency that makes a PE firm raise an eyebrow? It won't look like an agency at all. It'll look like a tech company with an unfair advantage in creative.

And it'll have figured out something the rest of the industry is still arguing about: the value was never in the Photoshop file.

It was in the thinking behind it. And in the system that gets it everywhere.