The AI services land grab: why your business model decides who will back your startup

Lawhive set out to sell software to law firms. The firms, mostly small high-street practices, were reluctant to buy. So Lawhive stopped selling to law firms and became one itself. Today, about 500 lawyers work through its platform and annual revenue has passed $35m, up sevenfold in a year. In February it raised a $60m Series B to take the model across the United States.

Read that as a simple pivot story and you will miss the point. Lawhive changed its customer, its business model and, without quite announcing it, the kind of investment capital it needed. The same choice is now arriving for every founder building with AI in a services market, from accounting and IT support to lettings and law.

There are only three ways to reach a customer an incumbent firm already serves. Sell the incumbent software. Compete with it for the client. Or buy it, client list and all.

Most founders treat this as a technology choice. It is not. It is a choice about how you acquire customers, and each route is paid for with a different kind of capital. Pick the route and you have picked your investors, whether you meant to or not.

What makes this urgent is timing. In February, PitchBook's analysts declared SaaS dead. Yet in the same quarter, global venture deal value in enterprise SaaS reached a record $173 billion. The money has not left software. But the conviction is already moving, and the investors leading that move do not agree on which road wins.

For founders, that is the trap. You have to commit to a business model before anyone knows which one will win. Then you have to pitch it to investors who are already sure they know.


Selling to the incumbent: the road that narrows

The first route is the familiar one. Build vertical software and sell it to the firms already doing the work. It still works where buyers are large and complex. It struggles where they are small and fragmented, which describes much of the services economy. The last generation of vertical SaaS built for small businesses largely stalled, and Lawhive's pivot is the same lesson told from the inside.

AI adds a second pressure. If software can do much of the work itself, the client may prefer to buy the finished work. Jake Saper, a general partner at Emergence Capital, argues that per-seat pricing is dying, and that a software company's chances of survival depend on how far it shifts towards selling outcomes.

So, the proof investors need from founders selling software to incumbents has changed. Showing that customers use the product is no longer enough. The founder must show why those customers will keep paying for the tool when a rival offers to do the job itself.


Competing for the customer: the road venture still recognises

The second route is to become an AI-native firm. Hire the lawyers, accountants or engineers, give them AI, and win clients from incumbents on speed and price. Because growth is built rather than bought, this route still looks like venture. Lawhive's Series B, per Fortune, drew Balderton, GV and Jigsaw, although it was led by Danaher co-founder Mitch Rales, an industrial operator rather than a venture fund.

It is harder than it looks. Saper says founders must effectively "build McKinsey and Stripe at the same time", and he looks for domain credibility in the early team, because a services company is selling itself.

The trap he warns about is mirage product-market fit: revenue growing fast and customers delighted, but humans still delivering most of the service. That is a low-margin services firm, and in his view it should not take venture capital. He wants a credible path to at least 70% gross margin.

So, the proof for this route is leverage: each lawyer, accountant or engineer producing more revenue every quarter.


Buying the customer: the road that changes your capital

The third route borrows private equity's oldest move. Buy fragmented local firms, put AI inside them, and own the clients outright. General Catalyst has committed $1.5 billion of its latest fund to incubating companies built to do exactly this.

London's Dwelly shows what the route costs. It buys independent letting agencies and runs them on a shared AI platform. Its Series B in July combined $95m of equity, co-led by EQT Growth and General Catalyst, with a $75m debt facility from Trinity Capital.

That mix is the point. Sahil Patwa, a co-founder of the Berlin roll-up fund Tenet, which launched with an €80m target, argues that the first acquisition is almost always funded entirely with equity. Debt follows once there is an operating history to lend against, and, as we argued in our August issue on venture debt, a lender is also weighing who stands behind you. Patwa expects 25 to 30% dilution per round.

The proof is specific too. Patwa says Series A investors want evidence that the founder can buy at sensible prices and lift profits afterwards. Dwelly says each of its property managers now handles upwards of 300 units, roughly triple a traditional manager's load. That is its own figure, not an audited one, but it is exactly the evidence the route demands. A roll-up is judged on its acquisitions, not its product demo.


Where the rules decide the roads

Regulated services add a test that comes before any investor conversation: which routes does the law allow? Legal services show how sharply the answer varies.

England and Wales let non-lawyers own law firms through alternative business structures licensed by the regulator. Scotland, just across the border, still does not. The Solicitors Regulation Authority, which oversees firms in England and Wales, has authorised Garfield.Law, an AI-driven firm pursuing small debt claims. And that openness to outside ownership let London-based Lawhive, a technology company, buy Woodstock Legal Services in what it called the first acquisition of a traditional firm by an AI platform.

In the United States, the legal AI company Eudia had to open its law firm in Arizona, the only state to have permanently changed its ownership rules. In most other states, a technology company still cannot own a law firm outright.

The lesson travels beyond law. In any licensed profession, check which roads exist before choosing one.


The investors have already chosen sides

Building an investor target list for an AI-powered services company might appear straightforward. Find the funds backing the category and start there. The trouble is that those funds disagree, openly, about which business model wins.

General Catalyst and Tenet are betting on buying. Saper is betting on building. He believes most venture money in the category has gone into roll-ups yet thinks starting from scratch is the likelier route to an enduring company. His Emergence partner Lotti Siniscalco calls the acquisition model the roll-up trap.

So, investors do not judge a pitch on its merits alone. They judge it against their own view of which business model will win. A roll-up pitched to a builder meets an objection settled before the meeting began, and the reverse is equally true. Founders already screen investors by stage and sector. Here they must also screen by conviction, which is why the partner matters more than the firm.


No shared yardstick

There is a second problem, and it is a consequence of how new this all is. A SaaS founder can assume the investor across the table knows what good looks like: recurring revenue, net retention, gross margins near 80%. An AI-native services founder cannot. They must explain the model and propose the measures it should be judged by, in the same meeting.

Saper's own approach shows what that looks like. He tracks three layers. First, a measure for each step of the service, such as how long a task takes with AI and a person together, tracked every quarter and owned by a named individual. Second, annualised revenue per service employee, showing each lawyer or accountant carrying more of the business over time. Lastly, gross margin, which he expects to be bumpy while the AI platform is still being built.

The lesson is that the metrics are part of the pitch. Arrive without them, and the investor will apply the yardstick they already know, which is usually the wrong one.


Choose the road, then the money

Before choosing a route, three questions are worth asking.

Which asset do you actually hold: a product strong enough to sell to incumbents, the domain credibility to compete with them as an AI-native firm, or access to the capital needed to buy them?

Can you show the proof your route demands, and have you defined the measures that prove it, whether that is customers paying for outcomes, rising revenue per person, or acquisitions that lift profits?
And does the investor across the table believe in your road, or in another one?

In this market, the route to the customer and the route to the capital are one decision. The founders who make it deliberately will seek out the investors who share their conviction. The rest will find out in the pitch meeting.

Let's talk.