OPINION | The rise of Vertical AI

Now, artificial intelligence is ushering in a similar, arguably even bigger shift. But this time, the transformation goes beyond digitising work. It’s about redefining what work even is.

The cloud era was about making human workflows digital. The AI era is about letting machines handle those workflows altogether.

Today, voice AI is already managing customer service calls, scheduling appointments, and qualifying sales leads. Generative agents are processing contracts and insurance claims without human review. AI is replacing entire layers of routine labour in the middle and back office.

This is why the AI opportunity is orders of magnitude larger than the cloud opportunity. Cloud vendors sold per-seat licenses. AI companies can sell outcomes: calls handled, claims processed, and sales closed. All without needing an army of human operators. The total addressable market shifts from “software spend” to “labour spend,” which is over 10 times larger in most industries.

Why the future will be vertical

While foundation model providers like OpenAI and Anthropic build the core, a lot of value will accrue at the application layer on top, with companies that are building for the needs of specific industries.

 I believe that the next decade will belong to Vertical AI: specialised applications that marry powerful models with domain-specific data, workflows, and go-to- market expertise.

 1) Integration is everything

Every industry runs on its own cocktail of software: CRMs, ERPs, scheduling tools, and compliance systems. These rarely talk to each other, and building seamless integrations is slow, technical, and unglamorous work. Horizontal AI platforms can’t possibly keep up with that complexity. Vertical players, however, can obsess over a single stack and build the deep integrations that make AI actually usable.

2) Industry workflows aren’t interchangeable

A home-services company doesn’t work like an insurer, and an auto dealership doesn’t work like a hospital. In home services, once a customer books a job, the system has to dispatch a technician, route them efficiently, and handle payments, all in real time. Building AI that fits into such rhythms takes deep domain empathy. Vertical startups can design around these nuances, while horizontal players must settle for abstractions.

3) Focused go-to-market beats brute force

Selling enterprise software isn’t about cold emails, it’s about trust. A salesperson who knows the language of construction, real estate, or logistics can connect with customers far more credibly than a generalist. Likewise, industry conferences, trade associations, and journals provide concentrated channels to win visibility. The narrower the target, the sharper the brand.

4) The data flywheel

In AI, data isn’t just fuel, it’s a moat. Once a company wins enough customers in one industry, it accumulates a proprietary dataset that no newcomer can replicate.

A voice AI platform built for logistics firms, for example, can fine-tune its models on thousands of dispatch calls to better handle noisy warehouses, driver slang, and industry jargon. Every new customer strengthens the moat.

Where Vertical AI will break out first

The earliest winners will likely emerge in industries where voice is central to day-to-day operations. Voice AI is already astonishingly human-like, and its adoption curve is accelerating. Think of sectors that live and breathe on phone calls: logistics, home services, auto dealerships, real estate, construction, and hospitality. In each, workers spend hours each day repeating the same conversations, and those conversations are now being automated, one by one.

The next decade of enterprise software

The AI stack will mirror the cloud stack: foundational model providers at the base, horizontal infrastructure in the middle, and vertical applications at the top. Furthermore, with everything deeply embedded in industry workflows, owning both the process and the data loop.

History offers a clue to how this plays out. The cloud era crowned focused specialists like Veeva, Toast, and Procore as billion-dollar category leaders. The AI era will do the same, but with even higher stakes. The companies that combine domain depth, proprietary data, and thoughtful human-AI collaboration will redefine how industries function.

The next Veeva or ServiceTitan won’t be built in the cloud; it’ll be built on intelligence.

(Arjun Gandhi is Vice President at Nexus Venture Partners.)

Views are personal and do not represent the stand of this publication.