I’ve spent most of my career in and around agencies, and there’s a basic piece of math that has never changed much.
You win a few clients. You get busy. You hire people.
Then you win more clients, get busy again and hire more people.
If you’re good at running the business, you can improve the margins around the edges. You can build better processes, offshore parts of the work, create templates, use freelancers and get smarter about utilization. I’ve done all of those things.
But ultimately, most service businesses have had a stubborn relationship between revenue and headcount. If you want to produce substantially more work, you generally need substantially more people to produce it.
That’s one of the reasons software companies have historically been so much more scalable—and often much more valuable—than agencies and consultancies.
I think AI has the potential to change that. But probably not in the way most agencies are currently using it.
Doing the same work faster isn’t a new business model
Most conversations I have with agencies about AI still start with efficiency.
How can we write the first draft faster? Can we automate reporting? Can AI help with research? Can one strategist do the work that used to require three people? Can we save ten hours a week on this account?
Those are all good questions. I’m asking them too.
But if all AI does is help an agency produce the exact same deliverables a little faster, we’ve improved the existing business. We haven’t fundamentally changed it.
We might get better margins. We might need fewer junior people. We might be able to take on another client without hiring.
Those things matter.
But the bigger opportunity is asking whether there are pieces of the service business that should stop behaving like traditional services altogether.
That’s a much more interesting question.
There’s a new model starting to emerge
You’ll hear a few different names for it. AI-native services. Services-as-Software. Productized services. Outcome-based businesses.
I care less about the label than the idea.
Instead of selling the customer people and hours to perform a service, you build a system that produces the outcome and sell the outcome itself.
Think about how traditional software works. A company sells you an accounting platform, and your people use it to do the accounting. They sell you marketing software, and your people use it to do the marketing. They sell you a CRM, and your people use it to do the selling.
The emerging model flips that around.
The customer doesn’t necessarily want another tool.
They want the work done.
That creates an interesting opportunity for people who already know how to do the work exceptionally well.
Agencies actually know something software companies often don’t
For years, agencies have looked at software businesses with some envy because of the economics.
But service businesses have an advantage that is easy to underestimate.
They know the workflow.
Not the sanitized version in a process document. The real version.
They know where clients change their minds. They know which inputs are always missing. They know which seemingly simple decisions require judgment. They know the exceptions that happen often enough that they’re not really exceptions anymore.
They also know what “good” looks like.
That matters tremendously when you start automating work.
A software developer looking from the outside may understand the steps in a process. The person who has done it 500 times understands where it breaks.
That is exactly the kind of expertise I wrote about recently when I said that experienced operators shouldn’t start by asking what AI company they should build. They should start with the problems they understand better than other people.
The same thinking applies inside an existing service business.
You may already have the raw material for a product hiding inside the work you perform every day.
The opportunity isn’t necessarily to build SaaS
This distinction is important.
For the last 15 years or so, whenever an agency owner said, “We should productize this,” the conversation usually drifted toward SaaS.
We should build a platform. We should create a dashboard. We should turn this internal tool into software and sell subscriptions.
Sometimes that works.
A lot of times it doesn’t.
The agency suddenly discovers that it has created an entirely different company requiring product management, engineering, customer support and a software sales motion it doesn’t understand.
I’ve seen plenty of businesses get distracted by that.
What’s different now is that you may not need to turn the customer into a software user at all.
You can use technology internally to change how you deliver the service while continuing to sell something the customer already understands.
They aren’t buying software.
They’re buying the result.
That may be a much easier transition for a service business because you don’t have to teach the customer a new behavior while simultaneously reinventing your own company.
Imagine the difference inside an agency
Take something like competitive intelligence.
The traditional version might involve an analyst spending 15 or 20 hours gathering information, reviewing competitors, looking for changes and turning all of it into a monthly report.
The AI-efficiency version says: let’s use AI so the analyst can do that in five hours.
That’s helpful.
But the more interesting version is asking whether you can build a system that continuously monitors the competitive environment, identifies meaningful changes, applies the agency’s methodology for interpreting them and delivers a useful intelligence product to the client.
Now you aren’t selling 15 hours of research. You’re selling competitive intelligence.
The distinction sounds subtle, but economically it’s enormous.
You could ask the same question about reputation monitoring, content development, customer research, media intelligence, SEO analysis, account-based prospecting, reporting or any number of things currently delivered as labor.
Not all of them should become automated products.
But some of them probably should.
I’d look for five things
If I were inside an agency trying to identify where this opportunity exists, I’d start with the work that is already highly repetitive.
Not repetitive in the sense that it requires no expertise. Quite the opposite. The sweet spot may be a service that requires expertise but applies that expertise to a workflow you encounter over and over.
I’d want the outcome to be fairly consistent even if the inputs change. I’d want a meaningful portion of the production work to follow patterns. And I’d want the human judgment to happen at important checkpoints rather than being required continuously through every step.
I’d also ask whether the customer actually cares how many hours went into producing the result.
In many cases, they don’t.
They want the leads. They want the research. They want the analysis. They want the campaign. They want the problem solved.
Finally, I’d look for something that gets better as you do more of it. Maybe you accumulate proprietary data. Maybe the workflow improves. Maybe your system learns from previous outcomes. Maybe your team becomes exceptionally good at identifying the small percentage of situations that require human intervention.
That’s when you start building something that looks less like capacity and more like an asset.
This changes the role of the people too
Whenever this topic comes up, it quickly turns into a conversation about eliminating jobs.
I don’t think that’s the most useful way to look at it.
The more interesting question is: What should your best people be doing?
I’ve spent enough years managing agencies to know that highly experienced people spend a surprising amount of time doing work that doesn’t require highly experienced people.
They chase information. They assemble reports. They reformat things. They prepare first drafts. They move information from one system to another.
If technology can absorb more of that production layer, the human role can move toward the things we actually hired talented people to do: judgment, ideas, relationships, problem-solving, taste and making the difficult calls.
That’s not removing expertise from the service.
It’s concentrating it.
And from an agency owner’s perspective, that creates another important possibility: the knowledge that used to live exclusively inside individuals can begin to live inside the company.
That has value well beyond efficiency.
It also forces a harder pricing conversation
If something used to take you 20 hours and now takes four, what should you charge?
If your answer is “four hours,” you’ve missed a pretty important part of the opportunity.
Clients don’t suddenly value an outcome less because you developed a better way to produce it.
In fact, if you can deliver it faster, more consistently and with better information, the outcome may be worth more.
This is one of the places where AI exposes the weakness of hourly pricing.
The billable hour ties your revenue to your inefficiency. If you become dramatically better at producing the work, you literally reduce what you’re allowed to charge.
That is a strange incentive.
Service businesses have talked about value pricing for decades. AI may finally force the issue because the relationship between effort and value is becoming too disconnected to ignore.
You don’t have to transform the whole company
One reason businesses struggle with AI is that every conversation turns into transformation.
We need an AI strategy. We need to become AI-first. We need to rethink the agency.
Maybe eventually.
I’d start much smaller.
Pick one service.
Preferably something you already sell, understand deeply and perform repeatedly.
Map how the work actually gets done. Figure out what requires real judgment and what doesn’t. Build the production system behind it. Keep humans where they create value. Remove them from the pieces where they don’t.
Then sell the outcome.
See whether customers like it.
See whether the economics change.
See whether quality holds up.
See what you learn after doing it 25 times.
That feels much more practical to me than a company-wide mandate to “use AI more.” It is also the kind of focused opportunity Buildside’s Build work is designed to validate before a service business overcommits to a new model.
This is another version of turning expertise into equity
When I started writing about turning expertise into equity, the basic idea was that knowledge trapped inside a person isn’t an asset yet.
You can make a great living selling what you know. But if the value requires your presence every time, you still own something that looks a lot like a job.
The same thing is true at the company level.
An agency can have enormous expertise and still sell all of it one hour at a time.
But once that expertise gets embedded into workflows, data, technology, processes and intellectual property that can repeatedly produce an outcome, something changes.
The company begins to own the way the work gets done.
That can improve margins.
It can create recurring revenue.
It can reduce founder and key-person dependency.
It can make the business easier to scale.
And, importantly, it can create something that another company might eventually want to buy.
Now we’re talking about equity.
There’s a question I think every service business should be asking
Not:
“How can we use AI?”
That question is too broad, and it usually leads to a collection of tools.
I’d ask:
“What part of what we currently sell as labor could become a system?”
That question forces you to look at the business differently.
Maybe the answer is nothing. There are still plenty of things where the human is the product, and there’s nothing wrong with that.
But if you’ve spent years doing the same valuable work for dozens or hundreds of customers, there’s a reasonable chance you’ve created more than a service.
You may have created a methodology.
A workflow.
A data set.
A way of making decisions.
A repeatable outcome.
AI may finally give you a practical way to turn those things into something the company owns rather than something the company repeatedly rebuilds from scratch.
That’s much more interesting to me than simply figuring out how to write the next deliverable faster.
Is there a system hiding inside the service you already sell?
Buildside works with agency owners, experienced operators and founders to identify what can become more than billable work, validate the opportunity and build the company, product or operating model around it.
Start a conversation ↗Brian Cross is a founder, operator, investor and embedded executive who has spent 25+ years building companies from inside the work.