Agency owners are hearing two things right now, and they seem to contradict each other.

AI is going to destroy the traditional agency model. And AI is going to make agencies dramatically more valuable.

Sometimes, conveniently, both arguments come from the same buyer. Your business deserves a lower valuation today because AI is creating disruption. Then, in the next breath, here is how much more valuable the business could become once they add AI.

So which is it?

The answer is both. And the gap between those two numbers, what an agency is worth today and what it could be worth transformed, is exactly where the AI agency rollup lives.

This isn't theoretical for me. I'm actively working on an AI-forward agency aggregation strategy: bring multiple agencies onto one cap table, centralize the investment in AI, and change how those businesses deliver work. Plenty of other buyers are looking at variations of the same idea.

So I want to break down three things. What an AI agency rollup actually is. How we got here. And what buyers and sellers should do about it.

What Is an AI Agency Rollup?

A rollup acquires multiple businesses and brings them onto the same cap table. Instead of owning an individual agency, the former owners now own part of a larger combined platform. The theory is that the combined company can create more value than the individual businesses could on their own.

Some of that value is traditional: centralized finance, shared leadership, better recruiting, cross-selling, less duplication, more sophisticated sales and marketing, access to capital.

The AI rollup adds a heavier operating thesis. Rather than asking every owner to independently figure out how AI should change their company, the platform centralizes that investment. In the model we've been developing, that means a shared AI program management office, an AI PMO, supporting an initial group of five to ten agencies.

The agencies keep their brands, client relationships, market positioning, and individual client workspaces. They don't all have to become one generic national agency. But underneath those brands, they start running on the same intelligence, infrastructure, operating standards, data, and AI capabilities.

Over time the platform can grow to 10 or 20 businesses. You start with cornerstone agencies that bring meaningful cash flow, management depth, strong client relationships, and credibility. Then you add complementary agencies around them. As the platform matures, later acquisitions can be funded differently, including with debt supported by the combined company's cash flow.

The rollover percentage should evolve too.

"An early owner who rolls 50% is accepting significantly more platform risk than a later seller joining a proven business. That early owner should have the opportunity to receive more of the upside."

Later add-ons may roll less because the model has already been proven and much of the early risk has been removed.

A rollup is not one transaction repeated 20 times. The economics should change as the platform changes.

What the Centralized AI Capability Actually Does

One of the assets behind this thesis is GrowthOS, an agency-owner-designed platform for managing and fulfilling services like SEO, websites, content, social media, and paid search. Think of the capabilities of a GoHighLevel or a SEMrush, combined with project management, delivery workflows, analytics, training, client personas, and role-based AI agents.

That last piece matters most.

Most companies currently use AI as a pile of individual tools. One employee uses ChatGPT. Someone else uses Claude. A third person pays for five different writing tools. The company now has more subscriptions, inconsistent processes, and no idea where its data is going.

"That is not an operating system."

Role-based agents work differently. You start by identifying a repeatable role or workflow. What information does that role need? What decisions does it make? What does it produce? What standards must it follow? When does a person need to review or approve the work? Then you build an agent around that defined responsibility.

The goal isn't to replace a department with one magical robot. It's to give each department, finance, legal, sales, marketing, operations, HR, technology, agents that handle specific, repeatable portions of the work with consistent context and controls.

Those same agents can support due diligence before a deal closes. A financial agent reviews quality of earnings and flags irregular expenses. A sales agent analyzes pipeline conversion, concentration, retention, and account growth. An operations agent maps the delivery process and finds where work gets stuck. A technology agent inventories the software stack, data access, security, and integration requirements. Each one classifies risk, identifies missing information, and produces questions for the seller. After closing, those same functional areas become the integration plan.

The output is not "AI says this is a good deal." It's a structured system that gives the human deal team better information and lets it move faster.

The Operating Targets

Here is why buyers are interested. The modeled targets we've been working with include:

• Reducing service-delivery time by 30% to 50%
• Reducing reporting hours by 70% to 90%
• Reducing labor costs within individual departments by 25% to 40%
• Eliminating 20% to 35% of unnecessary software spend
• Payback on the AI investment within three to nine months
• Moving agency margins from roughly 15% toward 40% within 30 to 60 days of implementation
• Potentially approaching 60% margins in certain offshore delivery models
• A total EBITDA improvement of roughly 25 to 40 percentage points over six to twelve months

These are targets, not promises. I would not underwrite every acquisition assuming the maximum result in every category.

"That is how you turn a compelling thesis into a catastrophic spreadsheet."

The point is that AI-driven value creation has to be measurable. It should show up in delivery time, reporting hours, employee capacity, software expenses, gross margin, EBITDA, client retention, or revenue growth. If none of those numbers move, the AI strategy didn't create enterprise value. It created another software expense.

And real margin improvement is not the same as shrinking the company. If you cut half the team and lose 40% of the clients, you haven't transformed the business. You've made a smaller one.

The goal is to increase capacity and profitability while protecting or growing revenue.

How Did We Get Here?

The agency market was suited to rollups long before generative AI showed up. It's large and fragmented. Thousands of founder-led agencies have good clients, capable people, and established reputations, but limited management infrastructure. Many are owner-dependent. Many sell time instead of outcomes. Many have inconsistent revenue, limited recurring contracts, and significant client concentration.

Those characteristics suppress valuations. They also create opportunities for capable buyers.

AI accelerated all of it. Clients expect work faster. Employees can produce more in fewer hours. Traditional deliverables are easier to replicate. Labor-intensive services are being commoditized. Hourly pricing is harder to defend. New competitors are being built around AI from day one.

The traditional agency is being forced to rethink its delivery model while still serving clients, managing employees, generating sales, and making payroll.

"It is closer to rebuilding the airplane while continuing to sell tickets."

Owners have another option, in theory. They can make the investment themselves. Hire the technical talent. Redesign delivery. Retrain the team. Develop new IP. Accept lower profits while the investment is being made. Take the execution risk and keep 100% of the upside.

For some owners, that's the right choice. But many don't have the capital, the capabilities, the appetite for disruption, or the several years it takes. They see the opportunity but don't want to fund and execute it alone.

That's the opening. The buyer brings capital, technology, leadership, and implementation. The seller brings the existing business, client relationships, team, expertise, and institutional knowledge.

The rollup puts both sides into the same economic outcome.

Understanding the Acquisition Economics

Some simplified math.

An agency produces $1 million of EBITDA. In its current form it's worth 4x, or $4 million. Under a 50% cash, 50% rollover structure, the seller receives $2 million in cash at closing and $2 million of equity in the combined platform.

The seller has taken meaningful money off the table. They've also reinvested half their proceeds. That rollover equity should be treated as a new investment decision.

If the platform eventually produces a 3x return on rolled equity, that $2 million becomes $6 million. Total proceeds: $8 million, twice the standalone value of the agency. At 5x, the rolled equity becomes $10 million and total potential value is $12 million.

That sounds compelling. It isn't guaranteed. The result depends on execution, debt, dilution, the preference stack, governance, the eventual exit valuation, and where the seller's equity actually sits.

"A seller who rolls $2 million is not receiving a vague promise of future upside. They are making a $2 million investment in the buyer, the management team, the other agencies, and the strategy."

That deserves the same diligence as any other multimillion-dollar investment.

Rollover equity is not a bonus attached to the purchase price. It's a second deal, and it should be evaluated like one.

Where the Uplift Comes From

There are two drivers: EBITDA improvement and the valuation multiple.

Take an agency at $10 million in revenue and a 15% EBITDA margin, so $1.5 million of EBITDA. At an 8x platform-level valuation, that's $12 million of enterprise value.

Now assume the operating platform, role-based agents, reduced reporting time, improved delivery, and consolidated software move the margin from 15% to 40%. The company now produces $4 million of EBITDA on the same revenue. Even if the multiple stays at 8x, enterprise value goes from $12 million to $32 million. If the quality of the business improves enough to move the platform multiple toward 10x or 12x, that's $40 million to $48 million.

That is the full modeled case, not something I'd automatically promise a seller, investor, or lender. But it shows why buyers are willing to fund the transformation.

The opportunity isn't simply multiple arbitrage. The biggest component should come from changing the underlying economics. The platform may then earn a higher multiple because it has better margins, less owner dependence, more predictable revenue, a repeatable operating system, centralized technology and data, proprietary workflows, greater scale, more diversified clients, and a clear path to continued acquisitions.

But that expansion has to be earned. Combining a group of average agencies does not automatically create an exceptional company.

A pile of agencies is not a platform.

What If the Seller Does It Themselves?

Consider the alternative. An owner independently invests $500,000 into AI, management, systems, and new capabilities. EBITDA goes from $1 million to $1.5 million. Stronger margins, reduced owner dependency, and a better growth profile move the multiple from 4x to 6x. The company is now worth $9 million.

Net of the $500,000, the owner has created roughly $4.5 million of additional value compared with selling for $4 million today. If they can actually achieve that, doing it independently may be more profitable.

But they have to fund the investment, absorb the disruption, and execute the plan.

"They cannot opt out of making the investment and simultaneously expect a buyer to pay them today for all the value the buyer hopes to create tomorrow."

That's where the entry arbitrage comes from. The buyer acquires the business on its current performance and risks, then has to invest additional capital, accept the transformation risk, and do the work.

The difference between entry valuation and future valuation is not free money. It has to pay for the investment, the risk, and the execution.

What Should Buyers Do?

The first rule: do not confuse an AI narrative with an AI operating plan. "We will add AI" is not an investment thesis.

You need to identify exactly where AI changes the economics of each business. Employee capacity? Delivery costs? Gross margin? Turnaround time? Reporting hours? Software spend? Recurring IP? The ability to sell outcomes instead of hours? Retention? A capability clients will pay more for?

If you can't connect the investment to revenue, margins, retention, capacity, or enterprise value, you don't have a transformation plan. You have a technology expense.

Buy the right raw materials. The most attractive acquisition may not be the agency already using the most AI. You may be better off with a company that has durable client relationships, strong domain expertise, proprietary or structured data, repeatable workflows, capable middle management, services that benefit from automation, and an owner willing to lead change. The question isn't "how advanced is this agency today?" It's "does this agency have assets worth transforming?" A badly positioned agency doesn't become attractive because you automate it. You may just help it deliver an unwanted service more efficiently.

Underwrite the implementation. Build the AI value-creation plan before finalizing the valuation. Map the existing workflow. Identify the agents to be deployed. Estimate implementation cost. Decide who leads the change. Identify affected employees. Set baseline metrics. Define success at 30, 60, 90, and 180 days. If you believe margins can go from 15% to 40%, show where each of those 25 points comes from: capacity, faster delivery, software consolidation, headcount, revenue growth.

"If the entire investment case is hidden inside one spreadsheet cell labeled 'AI efficiencies,' you have not completed the underwriting."

Diligence for the future operating model. Understand which services stay valuable, which get enhanced, and which get commoditized. Review client contracts, data rights, privacy obligations, IP ownership, and whether AI-generated work creates new risk. Identify who holds the critical knowledge. And determine whether the owner genuinely wants to lead the transformation. A seller who wants rollover economics without post-closing responsibility is not necessarily aligned.

Structure the rollover properly. The seller needs to know what entity they own, what percentage, what class of equity, what debt sits ahead of them, what preferences investors receive, how dilution works, how future acquisitions affect their ownership, what governance rights they have, what happens if they leave, and what creates liquidity. A headline valuation means very little if the rollover equity is buried under debt and preferred returns.

Underwrite a downside case. The deal should still work if the transformation takes longer than expected and shouldn't depend on multiple expansion. Model half the cost savings. Model flat revenue. Model margins improving over 18 months instead of six. Model never getting the 10x or 12x.

AI may improve the business, but it won't rescue a bad acquisition, fix poor retention, or turn a reluctant founder into a capable integration leader.

What Should Sellers Do?

Sellers have three broad choices.

Option one: make the transformation yourself. If you go this route, don't settle for adding a few tools and updating the website. Build evidence. Track gross margin, EBITDA margin, revenue per employee, time to complete work, reporting hours, client retention, delivery capacity, recurring revenue, software costs, owner dependency, and the share of work produced through repeatable systems. Buyers pay for demonstrated results far more readily than for an AI roadmap. If you can show delivery time fell 40%, reporting hours fell 80%, and margins rose without damaging revenue or retention, you've created something a buyer can underwrite.

Option two: sell the entire business today. This gives the most certainty. You take your proceeds and transfer the execution risk to the buyer. You also give up the upside. That may be exactly right if your priority is complete liquidity and a clean exit. Not every seller should roll equity. If you don't believe in the buyer, don't want to stay involved, or can't tolerate the risk, take the liquidity and price the deal accordingly.

Option three: take cash and roll meaningful equity. This is the middle ground. You reduce your exposure while keeping a stake in the future company. But don't focus only on the headline price. Diligence the buyer. Who owns the AI strategy? What have they implemented before? What capabilities already exist? How much capital is committed? What happens when the transformation temporarily reduces profitability? How does the platform decide which agencies get investment? How are conflicts between owners resolved? How does your rollover get diluted? What sits ahead of you? What has to happen before you see liquidity? What exactly is expected of you after closing?

And separate three economic components: the value of your business today, your compensation for working after closing, and the potential return on your rollover.

"Salary is payment for future labor. The purchase price is payment for the business you built. Rollover equity is an investment in the future platform."

Blur those three together and it becomes almost impossible to know what you're actually receiving.

Who Gets the Upside?

AI agency rollups aren't based on the idea that current owners have failed. They're based on the idea that the next stage of the market may require more capital, technology, leadership, and coordinated execution than many independent agencies want to provide on their own.

The buyer's opportunity isn't buying companies cheaply. It's building the infrastructure and operating capability that justify a higher future value. The seller's opportunity isn't negotiating the highest number at closing. It's deciding where they want to sit in the value creation that comes next.

Build it yourself. Sell and walk away. Or take some money off the table, stay invested, and help build a larger platform. There's no universally correct answer. But there is a correct way to evaluate the choice: understand what the business is worth today, what it costs to transform it, who is taking the risk, and whether the AI implementation is actually moving the numbers.

And make sure the future upside belongs to the people responsible for creating it.

Listen to the Full Conversation

Hear Peter Lang walk through the AI agency rollup thesis in full: why buyers are underwriting transformation rather than promising it, how the 50/50 rollover math works and where it breaks, what a role-based AI agent actually does inside an agency, and the questions every seller should ask before rolling a dollar of equity, on the Agency Acquisitions & Exits Podcast.

If you're an agency owner or buyer working through any version of this decision, join the Agency M&A Slack community or start the free 21-day Programmatic M&A Training.

‍