Boring is sexy: the anatomy of a good vertical AI startup

What separates a fleeting AI feature from a truly defensible AI company?

While everyone watches the foundation model labs race toward AGI (and burn through capital at a rate that would make even a crazy rich asian blush), some of the most compelling AI companies are being built in spaces you've probably never thought about.  Home services. Insurance brokerage. Distributors. Healthcare provider and payer operations. Loan origination. These were not always sectors that consistently make the cut at partner meetings on Sand Hill Road. And yet they are precisely the ones that are most interesting when it comes to applied AI. 

Amplify has been investing in vertical AI since way before it was cool, getting started back in 2014 with Enlitic. They were initially applying Deep Learning to radiology, and we discovered the hard way that the technology just wasn't quite there yet; the company had to go through a significant pivot. Today, the technology is unequivocally here. 

The rise of powerful, accessible foundation models has unlocked an entirely new playbook for technical founders to build category-defining applications in specific, complex non-tech industries.

The best vertical AI companies all follow the same four-part playbook. Take Avoca, who today announced $125M across their Series A and B that we're proud to have invested in:

• They're boring: The most exciting opportunities are in the least "sexy" markets
• They own the first mile: There's a major strategic imperative to own the first customer interaction
• They build a proprietary data asset that even the big AI labs can't replicate
• They become a system of action: They own a workflow, thus replacing the static "system of record" with a dynamic "system of action"

Let's dig into each of these.

Boring is sexy (& big)

Think about some of the largest sectors of the economy, like home services, healthcare, insurance, logistics, etc. These are massive, resilient pillars of how the world actually works. And yet, for years, they’ve been largely ignored by top technical talent.

Home services, where Avoca plays, is an essential layer for maintaining the country’s aging housing stock and represents a multi trillion-dollar market.

Similarly, insurance brokerage functions as the infrastructure layer that packages and distributes insurance products to businesses, with the market projected to exceed $140 billion in the U.S. alone by 2025.

These aren’t niche opportunities. They’re foundational markets that only feel boring because they lack the glamour of traditional tech or frontier AI research. But that perception is exactly what makes them interesting.

From an investor’s perspective, neglect creates opportunity. And right now, that neglect is setting up a perfect storm:

• Lower competition. These less crowded, “boring” verticals offer founders the chance to build with far less competition from well-funded teams compared to saturated spaces like horizontal sales and marketing tech.
• Stagnant incumbents. The existing software in these industries is often decades old and built on outdated tech stacks, leaving incumbents poorly equipped to attract top AI talent or move at the pace required to innovate.
• Desperate need for productivity. Many of these industries face a severe labor crunch, with aging workforces and rising reliance on expensive outsourced labor, making productivity gains not just desirable but essential for survival. The skilled trades are experiencing a massive retirement wave: five people leaving for every two that enter.

This is the landscape Avoca entered: a huge, underserved market with a desperate need for the productivity gains that only AI can unlock.

Own the first mile

The company that controls the initial point of customer contact wins. Agentic workflows require full context, with end-to-end access to data across the entire system. You cannot intelligently automate a workflow if you only see a thin slice of it. The goal is not just to participate in the workflow, it is instead to own the workflow’s front door.

In home services, 80 to 90% of revenue is booked over the phone. By owning that first interaction, Avoca starts capturing the full context that flows through the rest of the business, from dispatch to invoicing to customer lifetime value.

Codes Health, another Amplify company, also owns a critical entry point: the messy, fragmented process of gathering and structuring medical information in injury and malpractice legal cases. By controlling how that raw data gets ingested and contextualized, Codes defines the source of truth that shapes everything downstream, from case strategy to settlement. This position becomes even more powerful in an agentic world, where systems reasoning across workflows demand complete, structured context. Like Avoca, Codes isn't just participating in the workflow; it's capturing the moment where the most important data is created, positioning itself to expand from ingestion into orchestration and ultimately own the entire stack.

You must build a proprietary data asset

With the increasing commoditization of powerful foundation models, vertical AI companies must be constantly creating and curating a proprietary data asset. As models get better, the only defensibility comes from those who are able to build proprietary data aggregation at scale. 

The data must be unique. It must be context-rich. And it must compound over time — aka the more customers use your product, the better your data becomes, which makes your product smarter and stickier. More importantly—and this is the crucial insight—they're capturing the business logic of their customers. Every business operates with a set of unwritten rules and subjective decision-making processes that even the owner couldn't fully articulate if you put a gun to their head. 

Avoca fits squarely into this model by embedding deeply into home services workflows, capturing rich, real-time data on how operators actually run their businesses. Over time, this allows Avoca to learn and encode the nuanced, often unwritten decision-making logic of its customers, turning day-to-day operations into a compounding proprietary data asset that continuously improves the product.

Another Amplify company, Datalane, represents a more direct expression of this idea. They are building the go-to-market data layer for selling into local businesses, but not as a static database. Instead, they are constructing a real-time knowledge graph of offline business operations: who owns what, how franchise hierarchies are structured, when locations open and close, and how ownership changes over time. The result is a living dataset that improves with usage. 

From system of record to system of action

Think about the last generation of enterprise software giants like Salesforce, Service Now, or Workday: these companies became behemoths by establishing themselves as the central "system of record" (SOR). They were the database where all the important information lived. But with modern AI, a new opportunity many are calling a “system of action” has emerged.

We believe the most durable vertical AI companies must become systems of action. The path starts with owning a workflow and capturing the proprietary data that comes with it. The key is capturing not just the "what" (customer information) but the "how" (the business's operational logic). Especially in these prototypically boring industries, the "how" is where the real details live.

Once you've captured this logic, you can begin to automate it. Avoca starts with booking and scheduling, but that's just the beginning. With a deep understanding of how each business operates, they can expand to own every adjacent workflow: dispatching technicians, ordering parts, managing marketing, handling finances. They evolve from being a tool used by the business to becoming the central operating system for the business.

It's still early, but we believe there's a clear playbook emerging for vertical AI companies:

• Find a massive, unsexy market ignored by the broader tech world
• Insert yourself at the very first customer touchpoint
• Use that position to build a proprietary data asset that captures not just information, but unique business logic
• Leverage that data to evolve from a simple record-keeper into an intelligent system of action that owns the entire workflow

Avoca represents the perfect execution of this playbook (along with other Amplify companies like Codes Health and Datalane). We’re incredibly excited to announce their $125M in funding today and can’t wait to see what Apurva and Tyson are cooking.