What AI-Enabled Roll-Ups Actually Are

Part 1 of The Founder's Guide to AI-Enabled Roll-Ups

The premise is simple: service businesses have terrible margins because they scale with headcount. Every new client requires more people. More people mean more cost. Revenue grows, but profit stays flat.

AI changes that equation.

If you can automate 30-50% of the repetitive tasks that consume your workforce, you break the linear relationship between revenue and headcount. The same team can serve significantly more clients. Or you can serve the same clients with fewer people. Either way, margins expand dramatically.

That's the theory. The question is whether it works in practice, at scale, across different industries. This chapter explains the operating model, walks through what AI actually automates today, and examines the early results from portfolio companies that have deployed this strategy.

Basic Model: Services + AI = Software-Like Margins

Professional services businesses have a structural problem. An accounting firm, a call centre, an IT managed service provider—they all share the same constraint. When a new client signs up, you need more people to serve them. Revenue scales, but so does cost.

The numbers tell the story. Traditional call centres operate at 10-15% gross margins. Accounting firms typically run at 15-25% EBITDA. Property management companies often struggle to reach double digits. These aren't bad businesses. They generate reliable cash flow, serve real needs, and can be quite profitable in absolute terms. But they don't compound like software.

Software businesses have the opposite economics. Once the product is built, the marginal cost of serving an additional customer approaches zero. That's why SaaS companies can achieve 70-80% gross margins and trade at 20-90x EBITDA while services businesses trade at 5-10x.

AI roll-ups aim to shift service businesses toward software-like economics without becoming software businesses themselves. The approach: automate the repetitive, labour-intensive tasks that drive headcount, while keeping humans focused on the judgment calls and relationship management that AI can't handle.

Consider the math. A 50-person accounting firm handles 500 clients. Staff spend roughly 60% of their time on data entry, basic compliance checking, and report generation—tasks that are repetitive, rules-based, and high-volume. If AI can automate half of that work, the same 50 people can now handle 750 clients. Revenue grows 50%. Costs stay roughly flat. Margins transform.

General Catalyst, which has deployed $1.5 billion into this strategy, claims some portfolio companies are doubling EBITDA margins within 12 months of AI deployment. Dwelly, their UK property management platform, reports doubling EBITDA margins at agencies where their technology is fully deployed. These aren't incremental improvements. They represent a fundamentally different business model.

General Catalyst's "Creation Strategy"

General Catalyst didn't stumble into AI roll-ups. They developed a systematic framework they call the Creation Strategy, and it operates in a specific sequence that differs from how most investors approach service businesses.

Step 1: Map industries. The firm analysed over 70 service industries to identify where AI automation could have the greatest impact. They selected roughly 10 verticals where 30-70% of tasks could be automated with current technology. The targets share common characteristics: fragmented markets, labour-intensive operations, ageing ownership, predictable cash flows, and tasks that are repetitive enough for AI to handle but complex enough that simple RPA failed in the past.

Step 2: Build or incubate the AI platform first. This is the critical difference from traditional private equity. Before acquiring any services businesses, General Catalyst either builds or backs an AI-native software company in the target vertical. The software company develops automation tools, proves capabilities through pilot programs with existing service businesses, and creates integration playbooks.

Titan MSP illustrates this approach. General Catalyst backed the company from its first financing round, supporting the team as they built AI tools for managed service providers. Through pilot programs, Titan demonstrated it could automate 38% of typical MSP tasks. Only then did they acquire RFA, a well-established IT services firm serving financial services clients.

Step 3: Acquire distribution. Once the AI platform is proven, the company becomes an acquisition vehicle. It buys established services businesses with existing customers, recurring revenue, and trained workforces. These acquisitions provide distribution for the AI platform and immediate cash flow to fund further expansion.

Step 4: Deploy AI across acquisitions. Each acquired business gets integrated with the AI platform. The automation handles routine tasks while humans focus on higher-value work. As the platform processes more data across more businesses, the AI improves. Better AI enables more automation. More automation improves margins. The cycle compounds.

Step 5: Repeat. Improved margins generate cash flow for additional acquisitions. Each new acquisition benefits from an increasingly sophisticated AI infrastructure. The platform becomes more valuable as it scales.

This sequence matters because it inverts the traditional PE approach. PE firms typically buy first, then optimise. They look for operational improvements, cost synergies, and multiple arbitrage opportunities. The improvements are real but incremental.

General Catalyst builds the transformation engine before deploying capital to acquisitions. The AI capabilities are proven in pilots before being rolled out at scale. Integration playbooks are developed before the first deal closes. It's a higher upfront investment, but it positions the platform for margin transformation rather than margin optimisation.

The firm has stated that it's building permanent capital vehicles with no forced-exit timeline. This matters because the compounding effects take time to materialise. A 3-5 year PE hold period may not be long enough to realise the full value of margin transformation.

What AI Actually Automates Today

The marketing around AI roll-ups can get ahead of reality. It's worth being specific about what current AI capabilities actually accomplish versus what remains aspirational.

High-confidence automation (deployed and working):

Customer service interactions represent the clearest success story. Crescendo, General Catalyst's call centre platform, claims to achieve 80-90% automation for routine customer inquiries. When deployed at a regional telecom, Crescendo tripled the number of resolved calls in the first week by eliminating bottlenecks that had caused customers to stop reporting problems due to busy signals and long hold times.

Data entry and document processing work well. AI can extract information from invoices, contracts, and forms with high accuracy. This eliminates significant manual work in accounting, legal, and administrative functions.

Scheduling and coordination—booking appointments, coordinating open houses, managing calendars—can be largely automated. Dwelly's AI fully coordinates open houses and cuts repair wait times by 40%.

Basic compliance checking against known rules and thresholds is automatable. The AI flags exceptions for human review rather than replacing judgment entirely.

Report generation, particularly standardised financial or operational reports, can be automated once templates and data sources are established.

Emerging automation (working but earlier stage):

Early-stage reasoning tasks are improving rapidly. AI can now handle basic tax analysis, contract review, and recommendation engines. Eudia, General Catalyst's legal services platform, offers fixed-fee legal services to Fortune 100 clients, including Cargill, Del Monte, and Stripe—a pricing model that only works if AI substantially reduces the labour required per matter.

Predictive maintenance and operations forecasting show promise in IT services and property management contexts.

Still requires humans (for now):

Complex judgment calls that require weighing competing considerations, understanding context, or making decisions with incomplete information remain human work. Strategic advisory, novel problem-solving, and regulatory interpretation in ambiguous situations aren't yet automatable.

Relationship management—the trust-building, empathy, and nuanced communication that maintains client relationships—remains essential. AI can handle transactions; humans handle relationships.

The honest framing is that AI handles 60-80% of the work that's routine, repetitive, and rules-based. Humans focus on the 20-40% that requires judgment, creativity, and relationship skills. This isn't a replacement. It's an augmentation. But at this scale, augmentation transforms the economics of the business.

How Returns Are Generated

AI roll-up returns come from multiple sources, and understanding each one helps evaluate whether the strategy makes sense in specific contexts.

1. Margin expansion

The obvious driver. If a business runs at 12% EBITDA margins and AI automation can push that to 35%, the value creation is enormous. General Catalyst targets portfolio companies reaching 30-40% EBITDA margins—a level typically associated with software businesses, not services.

The maths compounds. A $10 million revenue business at 12% margins generates $1.2 million EBITDA. At 35% margins, that becomes $3.5 million. If both businesses trade at 8x EBITDA, the transformed business is worth nearly three times as much.

2. Capacity expansion

Same team, more clients. When AI handles routine work, existing staff can serve substantially more customers without proportional cost increases. This drives revenue growth without hiring, further expanding margins.

Crescendo's telecom deployment illustrated this. By automating routine calls, the same infrastructure could handle dramatically more volume. Customers who had stopped calling because of hold times started calling again. Revenue expanded while costs stayed relatively flat.

3. Capability expansion

AI enables services that were previously uneconomical to offer. An accounting firm can provide real-time financial dashboards and proactive advisory services instead of just annual reporting. A property manager can offer predictive maintenance alerts. These premium services command higher prices and create competitive differentiation.

4. Market expansion

When AI reduces the cost to serve, premium services become accessible to smaller clients. An accounting firm that previously needed $100,000 in fees to justify the partner attention required for strategic advisory might now offer similar services to clients paying $25,000. This opens new market segments.

5. Acquisition of multiple arbitrage

Services businesses typically trade at 5-10x EBITDA. Software businesses trade at 15-30x or higher for growth companies. If AI roll-ups can demonstrate software-like margins and growth profiles, they may eventually command higher multiples.

This is the most speculative return driver. Nobody has yet proven that acquirers will pay software multiples for transformed services businesses. The valuation gap could persist even if the margin transformation succeeds.

The "Rule of 60"

General Catalyst frames its targets using a twist on the SaaS industry's Rule of 40, which holds that a software company's revenue growth rate plus profit margin should equal at least 40%.

AI roll-ups target what GC calls "Rule of 60": 10-20% revenue growth combined with 30-40% EBITDA margins, totalling 50-60%. For context, traditional service businesses typically score 15-25% on this metric. Software companies generally hit 40-50%. The AI roll-up target exceeds even software benchmarks.

Whether these targets prove achievable at scale remains the open question. But the return framework is clear: margin transformation is the primary driver, with capacity, capability, and market expansion providing additional upside.

Case Study: Crescendo's Numbers

Abstract models are useful, but concrete examples are better. Crescendo, General Catalyst's call centre platform, provides the clearest case study with publicly available numbers.

Traditional call centre economics:

Call centres typically operate at 10-15% gross margins in mature markets. Labour represents 60-70% of costs. Revenue scales linearly with headcount—more calls require more agents. Staff turnover runs 30-45% annually, creating constant hiring and training costs. The business model works, but doesn't compound.

Crescendo's approach:

Crescendo built an AI platform that can fully automate 80%+ of customer interactions. The remaining staff handle complex issues that require human judgment or empathy. By flipping the ratio of automated to human-handled calls, the cost structure transforms.

The results:

Crescendo achieved a $500 million valuation after its Series C round in October 2024. The company reports gross margins of 60-65%—roughly four times the industry average. They're on track to exceed $100 million ARR by the end of 2025.

In October 2024, Crescendo acquired PartnerHero, adding 200+ customers to its platform. The acquisition provides distribution for their AI technology while generating immediate cash flow.

What it demonstrates:

Crescendo isn't hypothetical. It's running, profitable, and scaling. The margin transformation from 15% to 65% gross margins represents exactly the shift from services to software-like economics that the AI roll-up thesis predicts.

The caveats are real. Call centres may represent an unusually good fit for AI automation given the high volume of repetitive interactions. Results in accounting, legal services, or property management may differ. And we don't yet know how durable these margins will prove as competitors adopt similar technologies.

But as a proof point that the model can work in at least some contexts, Crescendo is compelling.

Why This Differs from Traditional PE

AI roll-ups sometimes get lumped together with traditional private equity roll-ups. The structures share surface similarities—acquiring multiple businesses in a fragmented industry, seeking operational improvements, and pursuing scale. But the strategies differ in fundamental ways.

Traditional PE roll-ups:

PE firms typically pursue cost reduction as the primary value driver. They centralise back-office functions, standardise operations, reduce headcount, and negotiate better vendor terms. Improvements come from operational efficiency within the existing business model.

Capital structures tend to be debt-heavy. Leverage amplifies returns but creates pressure to hit targets and reduces flexibility during downturns.

Exit timelines run 3-5 years. PE firms buy with a clear plan to sell to a larger PE firm, strategic acquirer, or the public markets.

Management teams are often replaced. PE firms bring in "operating partners" with experience executing playbooks across multiple portfolio companies.

Brand consolidation is common. Acquired businesses frequently get rebranded under a unified platform identity.

AI roll-ups:

AI roll-ups focus on revenue enhancement and margin transformation rather than cost-cutting. The goal is to create new value through technological capability, not extract value from existing operations.

Capital structures emphasise equity over debt. Long-term compounding requires flexibility, and the transformation thesis needs time to play out.

Exit timelines are indefinite. General Catalyst has explicitly stated they're building permanent capital vehicles with no forced exit. The strategy benefits from compounding, which means holding periods measured in decades rather than years.

Founders typically stay involved. The operational knowledge of acquired business owners remains valuable during transformation. Deal structures often include founder equity rollover and continued operational roles.

Local brands are preserved. Rather than consolidating into a monolithic platform, AI roll-ups tend to maintain local brand equity while providing centralised technology and back-office support.

The philosophical difference matters. PE extracts value from existing operations. AI roll-ups aim to create new value through transformation. PE optimises margins incrementally. AI roll-ups target step-function changes to the fundamental business model.

This distinction has implications for founders evaluating acquisition offers. A PE buyer likely plans to cut costs and flip the business. An AI roll-up buyer plans to transform the business and hold it indefinitely. The post-close experience will differ substantially.

What This Means

The AI roll-up model represents a genuine innovation in how investors approach service businesses. The combination of AI-native technology development, strategic acquisitions, and long-term compounding creates a playbook different from traditional PE.

The early results are promising. Crescendo's margin transformation, Dwelly's operational improvements, and Titan's automation metrics suggest the model works in at least some contexts.

But significant questions remain. Can margin transformation be sustained as AI tools become more widely available? Will acquirers pay software multiples for transformed services businesses? How will competitive dynamics evolve as more capital chases this strategy?

The next chapter examines who's deploying capital to AI roll-ups and how their strategies differ. Understanding the competitive landscape helps evaluate which approaches are most likely to succeed.

Continue to Chapter 2: The Money Behind the Movement →

Or return to: The Founder's Guide to AI-Enabled Roll-Ups (Hub)

Capital Founders OS is an educational platform for founders with $5M–$100M in assets. We focus on frameworks for thinking about wealth—so you can make better decisions.

If it's your first time here, don't forget to Subscribe.