The General Catalyst Behind $1.5 Billion of AI Roll-Ups
How one firm is deploying capital across six service verticals to test whether agentic AI can permanently rerate the economics of labor-intensive businesses.
On August 12, 2025, Titan, an AI holding company incubated by General Catalyst, announced $74 million in funding and its acquisition of RFA, an IT managed service provider serving over 400 clients in the financial services sector across multiple countries. Titan built an AI platform designed to automate routine IT support workflows, then used that capital to acquire RFA as its first operating platform, layering automation into an established client base.
The deal was not a one off. It was the sixth publicly disclosed execution of a thesis General Catalyst has been building since late 2023, when the firm allocated $1.5 billion from its fund specifically for what it calls AI enabled roll ups. The premise: identify fragmented, labor intensive services businesses where agentic AI can automate 30 to 70 percent of workflows. Build the AI platform first. Then acquire the services firms that have embedded client relationships and steady cash flows but lack the technology to transform themselves. Inject capital, deploy the automation, and re rate the margin profile from the 5 to 10 percent EBITDA typical of traditional services toward the 30 to 40 percent range associated with software companies.
General Catalyst is not alone. Thrive Capital launched a dedicated $1 billion vehicle in April 2025 and convinced OpenAI to take an equity stake and embed engineering teams inside its portfolio companies by December 2025. Lightspeed, 8VC, and Khosla Ventures have all entered the space. But General Catalyst has the broadest portfolio, the longest track record, and the most publicly disclosed results. That makes it the clearest lens for understanding whether this model works, and where it might break.
About QuantFi: We run finance & accounting departments for scaling businesses. We help you identify where risk sits in your model, which funding paths are viable, and how to present these choices clearly to your board.
The Math Behind the AI Rollup Thesis
The global services economy generates approximately $16 trillion in annual revenue, according to Federal Reserve Economic Data. The entire software market is estimated to be $1 trillion. Venture capital has historically ignored services businesses because their margins are too thin, their growth is too slow, and their operations are too people dependent. Private equity has participated, but primarily through cost cutting and financial engineering rather than structural margin transformation.
General Catalyst is making a different argument. Marc Bhargava, Managing Director and head of the firm’s Creation Strategy, described the thesis on the Cognitive Revolution podcast in August 2025: the firm mapped 70 service categories and identified 10 where current AI capabilities can automate 30 to 70 percent of the work. The automatable tasks fall into four categories: customer support and communication, data entry and processing, content generation (presentations, reports, emails), and early stage reasoning and analysis.
The financial logic follows directly. If labor represents 55 to 65 percent of a services firm’s cost structure and you automate 40 percent of that labor through AI, you do not necessarily eliminate headcount. You enable each person to handle two to three times more work, addressing the chronic labor shortages that constrain growth in industries like accounting, legal, property management, and IT services. Revenue grows because the same team can serve more clients. Costs grow more slowly because the marginal cost of AI augmented labor is lower than the marginal cost of hiring. EBITDA margins expand.
That is the theory. The portfolio is the lab test.
Inside the AI Rollup Portfolio
To understand why this thesis is so powerful, you must look at the actual workflow execution. These portfolio companies are not just buying software licenses. They are building proprietary AI platforms, acquiring established firms, and embedding that technology directly into daily operations to transform margins.
Crescendo in Contact Centers Valued at 500 million dollars after acquiring PartnerHero. AI resolves up to 90 percent of frontline tickets across 50 languages. By charging clients per resolved ticket rather than per hour, they report margins four times higher than traditional operators.
Long Lake in Property Management Raised 670 million and acquired 18 businesses. AI agents handle resident inquiries and draft board materials, driving productivity gains that allowed the firm to reach 100 million dollars in EBITDA in under two years.
Eudia in Legal Services Raised 105 million to attack the legacy billable hour model. The platform builds a continuously learning intelligence system to handle contract review and compliance, allowing corporate clients to drastically reduce outside legal spend.
Titan MSP in IT Services Acquired RFA to deploy autonomous agents that handle IT ticket triage and failure prediction. General Catalyst projects this automation can handle 30%+ of standard workflows, potentially tripling net margins.
Dwelly in UK Real Estate Raised 93 million to consolidate fragmented letting agencies. AI completely manages tenant verification and maintenance triage, dropping problem resolution times from 50 days to 20 days and doubling EBITDA margins wherever fully deployed.
Accrual in Accounting Launched with 75 million to automate tax preparation. AI agents read complex financial inputs and produce draft returns for human professional review, removing massive labor bottlenecks in a highly regulated industry.
Where the Rollup Thesis Breaks
The early results across these six companies are genuinely notable. Long Lake’s path to $100 million EBITDA in under two years is unusual by any standard. Crescendo’s 4x margin advantage over traditional BPO suggests the automation thesis holds in at least one vertical. Dwelly’s operational metrics, from maintenance resolution to tenant placement, demonstrate measurable improvement at the workflow level.
But services businesses have never traded like software companies for structural reasons, and those reasons have not disappeared.
Integration risk compounds with speed. General Catalyst’s portfolio companies are acquiring multiple firms per year, often before the previous acquisition is fully integrated. The Renovo Home Partners collapse, covered in a previous Capital & Clarity edition, demonstrated what happens when acquisition pace exceeds integration capacity. The same pattern is possible here, particularly in verticals like legal and property management where local relationships, regulatory environments, and operational nuances vary significantly across jurisdictions.
The employment question is uncomfortable but unavoidable. General Catalyst frames the thesis as enabling each person to handle two to three times more work, addressing labor shortages rather than eliminating jobs. That framing works in the near term. But when a platform can automate 50 to 70 percent of tasks in a vertical, the implied headcount reduction over a five to ten year horizon is significant. The political and social implications of that trajectory will eventually constrain how aggressively these platforms deploy automation, particularly in regulated industries.
Competition is intensifying. Thrive Capital’s $1 billion vehicle now has OpenAI engineers embedded inside portfolio companies. Traditional PE firms with decades of services industry experience are evaluating the same thesis. And big tech companies, whose AI models power many of these platforms, could choose to sell directly to services firms rather than through intermediaries. The moat for AI enabled roll ups is distribution and proprietary data, not the AI itself. If distribution can be acquired by anyone with capital, the competitive advantage is execution speed and operational depth.
Finally, these are still early stage bets. The largest portfolio company, Long Lake, is less than three years old. None have operated through a recession. The margin transformation numbers are reported by the firms themselves, not independently audited. The bear case is not that the technology fails. It is that the operational complexity of simultaneously running acquisitions, integrating heterogeneous businesses, building AI platforms, and managing client relationships across multiple geographies proves more difficult than the thesis implies.
The Capital Allocation Dilemma
Every PE backed services platform now faces a three part capital allocation question. Build AI internally, which requires technical talent most services firms do not have. Buy an AI enabled platform, paying the premium that early movers like Long Lake and Crescendo already command. Or get acquired by one, accepting the strategic logic that distribution (your client base) is the scarce asset, not the technology.
General Catalyst’s portfolio makes the math explicit. The firm reports that some portfolio companies are doubling EBITDA margins within 12 months of AI deployment. If those numbers hold at scale, the valuation implications are immediate. A services business trading at 8x EBITDA with an 8 percent margin is worth a fundamentally different multiple than the same business with a 25 percent margin, different cost structure, and recurring revenue characteristics that resemble software.
For CFOs sitting inside PE backed platforms, the operational question is equally pressing. These AI enabled roll ups are not competing on price. They are competing on capacity and speed. Crescendo’s outcome based pricing means clients pay for resolved tickets, not for agent hours. Dwelly’s AI generates 10 validated offers per property in three days. Eudia’s platform aims to replace the billable hour entirely. Traditional operators competing against these models are not fighting a cost war. They are fighting a structural efficiency gap that widens with every month of AI deployment.
The acquisition math also matters. General Catalyst’s portfolio companies are actively acquiring services firms at prices that reflect standalone economics: typically 60 to 70 percent cash at close, 30 percent founder equity rollover, with founders remaining operationally involved and local branding preserved. For aging founders in fragmented industries, this represents a competitive exit option that did not exist 18 months ago.
The Actual Takeaway
General Catalyst’s Creation portfolio is the most comprehensive public test of a thesis that will define the next decade of services industry M&A. The firm has deployed over $750 million across at least ten companies, with publicly disclosed results in six verticals. The pattern is deliberate and repeatable: build AI, acquire distribution, re rate margins.
The numbers are early but directionally significant. If even half of the reported automation rates and margin improvements prove durable at scale, the valuation framework for labor intensive services businesses will shift permanently. CFOs, sponsors, and operating partners who are not modeling this scenario into their strategic plans are underestimating the speed at which the competitive landscape is changing.
The risk is equally clear. Speed, complexity, unproven durability, and the structural differences between services and software businesses all represent legitimate constraints. This is not a proven model. It is a well capitalized hypothesis being tested simultaneously across six verticals by one of the largest venture firms in the world.
That makes it worth watching closely.
How We QuantFi It
QuantFi works with PE sponsors, finance teams, and CEO’s running exactly the kind of services platforms that General Catalyst’s portfolio is acquiring or competing against. We sit between strategy and execution on AI implementation in the finance department: mapping finance/accounting workflows, implementing AI automations where relevant, and presenting the ROI behind the investments. In practice, that usually looks like three things:
Running focused AI sprints and workflow mapping to identify where automation can safely absorb 30–50 percent of current activity without breaking controls or client experience.
Redesigning finance and accounting operating models around those changes (roles, processes, KPIs, and reporting) so the savings and capacity gains actually show up in the numbers.
Building a three-scenario model (baseline, AI-augmented, and “platform/acquirer” case) that quantifies ROI, payback period, and valuation impact, then packaging that into a plan a CFO can take to their board and sponsors.
If your portfolio includes labor-intensive services businesses and you have not yet mapped workflows, sized the automation opportunity, and tied it to a concrete ROI case, that work is already overdue. The firms moving fastest are not waiting for external proof; they are generating it through controlled experiments and acquisition
Thanks for reading Capital & Clarity By QuantFi!
Subscribe for free to receive new posts and support our work.
thank you for the deep dive, we see fragments of this story across the vc newsletters we track every week but nobody had pulled it together like this.