What is an AI roll-up?
An AI roll-up buys services firms that do the same work and rebuilds their delivery around AI. It pays off only if each firm gets cheaper faster than the last.
An AI roll-up, often written "AI rollup", buys several services firms that do the same kind of work and moves all of them onto one AI-based way of delivering it. Its case rests on one question: how fast does each newly acquired firm reach a falling cost of work, and does that get faster with every firm added?
What decides whether an AI roll-up pays off?
Buying firms adds revenue. It does not, by itself, change what the work costs. The value is created after the purchase, when a firm's delivery moves onto systems where AI executes the routine and specialists own the judgment calls. Whether that rebuild works at all is covered in Does buying a services business and adding AI actually work? This note is about what comes next: doing it again, firm after firm.
The useful measure is time to the falling curve: the time from closing a purchase to the point where the human time behind the same job starts falling and keeps falling. In a strategy that works, that time shortens with every firm. The first rebuild is slow because everything is new. Later ones should be faster because much of the work is already done.
What makes each firm faster to rebuild than the last?
Three things have to carry from one firm to the next:
- A shared engine. The systems that run the routine steps are built once and reused. A new firm plugs into them instead of building its own.
- Playbooks. The way a firm is moved over is written down and refined each time: how its jobs are mapped, which steps move to the systems first and how its specialists are brought in.
- Shared judgment. Every judgment a specialist makes is captured with its reason and made available to every firm. A new firm starts with the decisions earlier firms already made. This is the data flywheel working across firms, not just within one.
If any of the three stays inside a single firm, each acquisition becomes a fresh project, and the strategy is a set of separate rebuilds rather than one that compounds.
What has to be true for that to repeat?
- The firms do the same kind of work. Lessons transfer only when jobs look alike. Buying firms that do different work resets the curve each time.
- Specialists stay and own the judgment. They are the source of what gets captured. A firm whose specialists leave adds jobs but little to learn from.
- Delivery runs one way. Firms keep their customers and relationships, but the work runs on the shared engine. Local variations of the systems split the learning.
- The gain stays in the business. Pricing by the result keeps the falling cost inside the business, as covered in What is outcome-based pricing?
Where does repeatability break down?
- Buying outpaces rebuilding. Moving many firms over at once is a management problem before it is a technology problem. Buying faster than firms can be moved stacks up firms still working the old way.
- Each firm is treated as a one-off. Custom systems for each firm mean nothing carries over, and the time to the falling curve never shortens.
- The case rests on the label, not the cost. If the plan works only when the combined business is valued like a technology company, rather than because the work keeps getting cheaper, it is fragile.
How can you tell if an AI roll-up is compounding?
Compare firms by when they joined. If each new firm reaches a falling cost of work sooner than the one before, and the cost keeps falling once it gets there, the engine, playbooks and captured judgment are doing their job. If every firm takes as long as the first, the strategy is buying firms, not building a repeatable way to run them.
Questions people ask
Is an AI roll-up the same as a private equity roll-up?
Not quite. Both buy many firms in one industry. The difference is the thesis: a classic roll-up wins on scale and structure, while an AI roll-up is built on changing how the work is delivered and making that change faster with each firm.
What industries do AI roll-ups target?
They look for work that repeats for similar customers, with results that can be checked. Accounting and tax, IT managed services, insurance brokerage, property management and back-office support are common examples.
Is an AI roll-up the same as an AI-native services company?
No. An AI roll-up is a way of assembling a business through acquisitions. AI-native describes how the work is delivered. A roll-up becomes AI-native only if the work itself is rebuilt around AI.
Questions or ideas? Email hello@skywakelabs.com.
Buyers, investors and partners: hello@skywakelabs.com
More notes
What is agentic BPO?
Outsourcing where AI agents work most cases and specialists take the exceptions. Real agentic BPO prices the finished case, not seats or agents.
ReadWill AI replace procurement teams?
Not the team. AI takes the paperwork of buying; people keep the decisions with no written answer: which supplier, on what terms and whether to buy at all.
Read