AI is Solving SaaS Product Adoption

How Vertical AI is breaking down traditional industry barriers

In our last essay, “Emerging Playbooks in Vertical AI,” we discussed how LLMs are enabling new vertical software strategies to automate work-to-be-done and—in the process—laying the groundwork for a new generation of systems of record. These developments are already changing the way that founders, product managers, and investors think about software architecture and roadmap. Vertical AI, however, is ushering in something far greater than novel product strategy.

Vertical AI is breaking down traditional barriers to industry adoption that have held back entire generations of SaaS startups. Past vertical platforms have achieved higher-than-average penetration rates than their horizontal peers, keeping switching costs high in environments lower-competition segments. Paired with varying budget and market education challenges, certain verticals seemed impossible ground for SaaS. AI, however, is transforming the traditional vertical go-to-market. LLM-based wedge products have proven the ability to drastically reduce market education, avoid rip-and-replace legacy systems of record, and change buying calculus altogether, with innovative new value propositions and pricing schema.

The early adoption curves of top-tier Vertical AI—some with growth rates that look more like best-in-class consumer or marketplace businesses than B2B software—make it clear that there is something unique to the formula. And if this phenomenon proves replicable throughout the economy, the already massive opportunity in Vertical is set to experience a generational acceleration.

One big question remains, however: if there are formulas to success in Vertical AI product and go-to-market strategy, what are its key components? What transferable learnings can founders and investors glean from early Vertical AI breakouts? In today’s essay, we will share our view on how Vertical AI is successfully breaking down adoption barriers, taking a first step in articulating what success looks like in this fast-evolving space.

1. Matching the Medium

No matter how product-led or well-designed a software solution is, it has traditionally required some level of user training. If a customer’s job is to parse Bills of Lading on behalf of a supply chain company, and that company adopts a new TMS to manage these documents, they must meet that solution in the middle. I had a particular way of doing business before; the software has a tailored architecture and UI; my new workflow will land somewhere in between. That adaptation process inherently takes time and money, mostly through workforce retraining. Historical software adoption, therefore, was an investment like any other: buyers take a risk that the long-term ROI of the product will outpace inevitable upfront costs (beyond just the purchase price).

LLMs give vertical software solutions the opportunity to greatly reduce the upfront investment cost of adoption. For example, in that same supply chain scenario, a new solution might parse the documents, send the results and request QA via email, and handle input and output of results from existing systems of record automatically. All the user has to do is focus on their core job—making sure everything looks right with the BoL and handling the edge cases—with no investment into learning a new UI required. Especially in traditional industries, where aversion to deviating from “how things have always been done” can be illogically high, reducing perceived upfront investment is meaningful.

Abridge is an excellent example of the power of reduced friction with Vertical AI. Physician communities within large medical systems are not typically breeding grounds for viral adoption. Founder Shiv Rao described how his vision was inspired by his experience as an attending physician: “I would pick up the phone after a procedure, like a Holter monitor or an echocardiogram, and I would start dictating the procedure report. People in the basement of the hospital were actively, synchronously on the line listening and typing, and then that report would end up in the medical record.”¹ He would then double down on matching the customer’s medium by embedding with their most common EHR, Epic.²

Vertical AI can not only meet customers where they work, but it can also enable client enterprises to meet their customers where they work. An HVAC business owner, for example, might struggle with last-minute appointment cancellations. There’s a complex web of communication between the customer, their technician, and central dispatchers. Along this web, there are many points of failure. LLMs can enable not only instant communication but also, in many cases, instant resolution—as in the case of rescheduling that client to a new slot and informing the technician. We see similar dynamics with Euclid portfolio companies TheraDriver in the ABA space and Indie Health in OT: Clinics not only save time and cost through AI, but they also drive a superior, more natural experience for their patients.

The point is that, unlike many traditional software solutions, Vertical AI is significantly more agnostic to UI form-factor, allowing it to meet a customer in their native workflow. Consider that LLMs’ superpowers center around natural language—conveniently also the primary medium for doing business! As Wayne Hu at SignalFire said, “Vertical AI can tackle previously intractable problems by addressing unstructured data, which makes up about 80% of all data and has been challenging for traditional SaaS to handle effectively.”³ So, the possibilities for wedge product form factors are many: texts, calls, emails, document creation, data entry, note taking, data summarization, needs matching, high-level analysis, training, learning, capturing tribal knowledge, and general communication. The surface area for Vertical AI, therefore, as it seeks to match the medium of every business across the economy, is limitless.

2. Cost-Effective Interoperability

The second primary reason vertical AI erodes traditional barriers to industry software adoption is its flexibility in integration. The ability of vertical software to achieve high stickiness is well known. Many public vertical software companies have been market leaders for 20, 30, and even 40 years. In some cases, there are industry dynamics that allow this, as in the case of GovTech players like Tyler Technologies or Granicus. In many cases, however, we would attribute this to the successful lock-in of many stakeholders throughout that vertical ecosystem. Good examples might be Yardi in real estate property management and Procore in construction ERP.

The universe of old, sticky, $100M- to $1B-revenue vertical software point solutions, however, is immense. Examples that come to mind include ECI Spruce (the predominant ERP for building material suppliers), Dentrix (the dominant practice management system for dentists), and many others. For emergent vertical SaaS businesses in these spaces—even ones that aren’t directly competing with these core systems of record to start—this can present significant headwinds. Monopolistic ERPs tend to throw their weight around, either to maintain leverage over emerging competitors or to tax them at the least. Epic, an 800-pound gorilla healthcare ERP with nearly 50% market share, is a perfect example. It would be very difficult for a point solution that signs Epic users to succeed without integration. This may be costly regarding the integration fees Epic charges, but it may also be logistically prohibitive.

While one of the major benefits of cloud SaaS was its multi-tenancy, the older a platform is, the more likely that its various enterprise instances are highly heterogeneous, with significant deviations in object architecture and taxonomy between customers. Achieving a successful enterprise integration with Epic EHR is one thing; building a software solution that can scalably manage these multivariate differences between customer schemas is another. Because data objects and their labels have natural language elements, LLMs can handle that disambiguation like the best of Epic power users but at orders of magnitude higher speed and accuracy.

Our friends at Bessemer Venture Partners articulated this integration point well. “To mitigate the risk of LLM commoditization, the best vertical AI applications will tackle an end-to-end workflow and be tightly integrated with existing systems—regardless of whether the application is an end-to-end or single-point solution.”⁴ While we agree with both points, early wedge products do not necessarily need to be formally integrated. We have seen several startups launch quickly into the market by building agents automating repetitive read-write tasks on behalf of a credentialed user—it’s not a long-term solution, but it’s a fast start that would have been near-impossible pre-LLM.

3. Agile Value Alignment

Another key aspect of how vertical software breaks down industry barriers concerns differences in how it can be bought and sold. There are two key components of this. First, the value prop and commensurate sales positioning. Second, the form factor and TCO are implied by pricing.

First, let’s take a step back into what vertical AI excels at. As mentioned, thanks to LLMs, anything related to natural language is a superpower. But it’s not just the ability to interpret or create language documentation; it’s also the scale at which LLMs can form conversations or interactions—or customer intake or document creation. These activities compose a large percentage of labor falling under the general and administrative bucket. Sometimes, that labor is in the form of in-house personnel; other times, it is outsourced to third-party services. Either way, vertical AI does not need to sell into a technology budget but can capture discrete OpEx spend.

This can be particularly germane to decreasing friction of initial adoption because most enterprises already have some form of overflow contingency. For example, a business might contract with a third-party call center operator to handle off-hours calls, a third-party medical coder to handle overflow in some instances, or virtual admins to supplement in-house staff. A vertical AI platform with a discrete services replacement wedge can sell into this preexisting channel. Because the output is the same—regardless of whether the counterparty is a vertical AI startup or a traditional third-party outsourcing shop—an executive has to take no additional steps in procurement.

Moreover, the LLM-powered solution can offer pricing that reduces the risk to positive ROI. Because vertical AI solutions are generally less UI-heavy, they may get to market with much lighter initial development investment. The more important ongoing cost consideration for the startup might be the burden of ongoing inference in addition to traditional cloud infrastructure. Because the price point of its solution does not have to subsidize the heavy initial development phase, however, the vertical AI startup can often price both under market—giving it clear appeal versus traditional services alternatives—and on a scaling factor that is directly tied to value to the customer. That said, when its value is anchored in service/labor costs, Vertical AI may still have a significant pricing advantage vs. traditional SaaS. As Alex Niehenke at Scale so aptly put it:

This change in delivery of value allows the vertical software vendor to price based on value rather than relative to employee salary, and the net result is profound. We have seen examples of where the vertical AI vendor is capturing 25% or 50% or more of an employee salary. This would suggest vertical markets will be five to ten times larger through the introduction of artificial intelligence.⁵

Even a services firm likely has to staff up and provide some training to onboard a client. An LLM-powered solution, however, can look much more like a self-serve PLG offering in that it monetizes purely on a usage basis, with a lighter initial penalty for spinning up and down. In the long term, this is potentially a negative path for vertical AI startups. Ultimately, the startup will want to adopt some of the favorable elements of the SaaS subscription model, and we find, in fact, that in enterprise environments, customers also appreciate the visibility of knowing what they will pay. It is somewhat trivial, however, to transition from a usage-based pricing model to a volume-tiered recurring contract if the product delivers consistent value to the customer. As is a common thread throughout our thinking on vertical software, expanding beyond the initial wedge always simplifies the pricing transition.

There’s much more we could say about vertical AI and pricing here. We will be sharing a future essay dedicated to this topic to unpack it in full. In sum, for several reasons—primarily related to new types of wedges that LLMs enable and how buyers perceive their value—vertical AI businesses are quickly circumventing many prominent barriers to adoption. As has always been the case in the technology world, these advantages of Vertical AI will not be universal and must suit the unique dynamics of each industry. “Easy to build,” as has always been the case in technology, can mean “easy to replace.” As we’ve discussed at length in past essays, Vertical AI must still look to workflow and data for long-term defensibility—but an earlier path to hyper-growth is rarely an unwelcome start.

4. Vertical Virality

Once a Vertical AI solution does break the seal on initial industry adoption, it benefits from the very same viral, network-based adoption effects as its successful SaaS peers. Vertical platforms generally benefit from catering to not only the core ICP, but also the relevant peripheral industry stakeholders. AppFolio appeals to owners and property managers. ProCore appeals to general contractors, subs, and owners. As alluded to above, Abridge appealed to clinicians, patients, and operators of keystone health systems like UPMC.

Once the benefits are widely proven—especially at an industry-leading organization—the enablers of fast scaling above kick in and growth can be breathtaking. Abridge grew from single-digit ARR to nearly $100M ARR in little more than a year. EvolutionIQ in the insurance claims processing space grew revenue 3x for 3+ years in the early days.⁶ Harvey AI is an excellent example of the effect Vertical AI can achieve in software-hold-out sectors once ROI is known and the referral and reference-buying network effect kicks in:

Harvey’s first major breakthrough came in early 2023 when Allen & Overy, one of the world’s largest law firms, announced it was rolling out the tool to its 3,500 attorneys. This was a watershed moment. Until then, AI adoption in major law firms had been slow, with many viewing these tools as either too risky or too simplistic for complex legal work… by 2024, the startup had expanded its client base from 40 firms to 235 across 42 countries.⁷

The step-change in user experience, savings, or revenue lift is so stark that even enterprise-focused Vertical AI businesses can achieve consumer-like growth. Early data from BVP’s portfolio companies saw average Vertical AI revenue growth of ~400% year-over-year⁸—extremely impressive for businesses with a likely mean stage of Series A+. We have confidence that, in aggregate, these new product strategies, G2M postures, and growth potentials are contributing to an overall business profile that looks very different from traditional Vertical SaaS: in building, in scaling, and ultimately, in capital needs. We will follow up with additional essays on these points: what the best look like and how that may fundamentally change capital needs.

The elements of the Vertical AI formula driving incredible success are by no means set in stone—we are still in the early days of this cycle. There will be much more to come on this topic. In the meantime, we hope this rundown of the key primitives to Vertical AI’s barrier-smashing success is helpful to founders considering the next generation of vertical platforms.

Thanks for reading Euclid Insights! If you know a Vertical AI founder thinking through their next idea, market, or wedge product, please reach out via LinkedIn, email, or Substack.

1. HIStalk (2023). HIStalk Interviews Shivdev Rao, MD, CEO, Abridge. HIStalk.

2. Abridge (2024). Abridge Becomes Epic’s First PAL, Bringing Generative AI to More Providers and Patients. Abridge.

3. Hu (2025). Frameworks for AI Vertical SaaS. SignalFire.

4. Bessemer Venture Partners (2025). Part IV: Ten principles for building strong vertical AI businesses. Bessemer Venture Partners.

5. Scale Venture Partners (2025). The future of AI is vertical. Scale Venture Partners.

6. EvolutionIQ (2023). Insurance AI Leader EvolutionIQ's Funding Total Rises to $33.1M With Series B Round While Valuation Passes $200M. EvolutionIQ.

7. Mathews (2025). Harvey AI came out of nowhere and took over legal tech. AIM Research.

8. Bessemer Venture Partners (2024). Part I: The Future of AI is Vertical. Bessemer Venture Partners.