Vertical AI①: The Future of AI is Vertical
In recent years, the world of SaaS (Software as a Service) has seen remarkable growth in Vertical SaaS, which specializes in specific industries and sectors. For example, software optimized for specific industries has emerged one after another, such as Procore for the construction industry, Toast for the restaurant industry, and Shopify for e-commerce. These companies have created immense market value while simultaneously revolutionizing how their respective industries work.
Article:
However, as the next step for Vertical SaaS, which grew by leveraging cloud computing and mobile technology, Vertical AI is expected to have an even greater impact. This is a new category of software that applies LLMs (Large Language Models) and generative AI specifically to vertical domains, and it possesses the following characteristics:
The ability to automate language-based repetitive tasks
Opening up business areas that traditional software could not reach by having AI perform tasks instead of (or in support of) humans
Applicable not only to core business workflows but also widely to support tasks such as back-office operations
In this article, we will explain step-by-step the background behind the emergence of Vertical AI, the vast market opportunities hidden within it, and specific examples.
1. The Emergence and Background of Vertical AI
1-1. A New Horizon Seen from the Success of Vertical SaaS
A while ago, software specialized for specific industries—so-called Vertical SaaS—was often considered "niche and boring." However, Mindbody, Shopify, and Procore have grown significantly, and today, the combined market capitalization of the top 20 publicly traded Vertical SaaS companies in the U.S. alone has reached approximately $300 billion. With this track record, the market potential of Vertical SaaS is no longer in doubt.
The next wave attracting attention is Vertical AI, which utilizes LLM-native applications. In 2023, new AI startups emerged one after another in fields with many language-based repetitive tasks, such as law, healthcare, and finance. It is expected that Vertical AI will be able to penetrate areas that traditional Vertical SaaS could not fully capture, or areas that were difficult to software-ize in the first place.
1-2. Background of Market Expansion and Differences from Existing SaaS
One of the reasons Vertical AI is receiving so much attention is the sheer size of its market. For example, according to U.S. government statistics, software-related spending accounts for only 1% of U.S. GDP, but the business and professional services sector, which involves many language-based repetitive tasks, is said to account for 13% of GDP. This suggests that by introducing AI into the "vast majority of areas that have not been software-ized," market expansion of more than 10 times can be expected.
Furthermore, Vertical AI can not only "replace existing legacy software" but also "provide solutions to areas where software had not been introduced at all." Traditional software products had high implementation and learning costs, making it difficult to persuade industries that were cautious about making adoption decisions. However, if the value provided by AI is significant, even large companies will actively consider its adoption.
2. Why Vertical AI Now?
2-1. High Growth of Emerging Companies and Acquisition Cases
In recent years, according to reports and portfolio analysis by investors such as Bessemer, there is data showing that LLM-native companies founded since 2019 boast high Annual Contract Values (ACV), annual growth rates exceeding 400%, and maintain healthy gross margins of 65%. In fact, exits through M&A have already occurred, such as CaseText being acquired by Thomson Reuters for $650 million, and Lexion being acquired by DocuSign for $165 million.
Following this trend, industry leaders are also beginning to actively incorporate AI, choosing between "buying or building AI capabilities in-house." For example, there are moves such as the following:
Thomson Reuters' acquisition of CaseText
DocuSign' acquisition of Lexion
AI integration in existing SaaS such as IntercomZapier, and Canva
The demand for Vertical AI is already being recognized not as a "future possibility" but as a "current business opportunity."
2-2. Creating New Workflows and Expanding Value
One of the characteristics of Vertical AI is that "the more an industry has repetitive tasks that can be replaced by AI, the greater the benefits." This is because analog processes and tasks that were previously performed manually can be dramatically streamlined through the introduction of AI.
For example, a service called EvenUp automates the creation of "Demand Letters" for law firms handling personal injury cases. By automating document creation, which previously took human lawyers and paralegals a significant amount of time, firms can reduce costs while increasing their client base. Such examples are beginning to be seen not only in the legal industry but also in a wide range of sectors such as healthcare, finance, and construction.
3. Core Workflows and Support Workflows
The functions provided by Vertical AI can be broadly categorized into those deeply rooted in an industry's "core business" and those that streamline "peripheral tasks (support tasks)."
3-1. Utilization in Core Workflows
A prime example of automating and streamlining core tasks is Fieldguide, which supports audit tasks with AI. The work of auditors meticulously checking vast amounts of documents and data is extremely time-consuming, but productivity is significantly improved by having AI take over data processing.
However, the more core the task, the greater the psychological and cultural hurdle of "do we really want to leave this to AI?" For example, while creating presentation materials for an investment bank can be automated, there would likely be resistance to leaving business negotiations with clients entirely to AI. Precisely because it is a core workflow, there is a tendency to be more cautious when introducing it.
3-2. Utilization in Support Workflows
On the other hand, support tasks often face less resistance to AI adoption than core tasks, and the productivity improvement effect from their introduction is significant. For example, hospital appointments, medical record entry, and back-office processing in financial institutions contain many tasks that can be performed by non-specialists. Abridge, which takes over the "so-called paperwork" for doctors, and the medical knowledge search platform ClinicalKey AI are prime examples.
However, there are an increasing number of cases where large horizontal SaaS providers for support tasks have already incorporated AI features, so for startups to enter, "how much industry-specific knowledge and data can be leveraged" becomes a key differentiator. For example, a mechanism that "automatically detects maintenance needs for solar panels and assigns the optimal technician" requires industry-specific know-how that is difficult to realize with general LLMs.
4. What is the Defensibility (Moat) of Vertical AI?
The criticism that "AI applications are nothing more than wrappers for LLMs" may indeed apply to some cases. However, there are many cases where industry-specific AI companies have established unique defensive moats by building data, product depth, and actual economic value.
For example,
Vertically specialized datasets
Product design that complies with legal regulations and security requirements
Long-standing networks with industry stakeholders
These elements form a powerful moat that is not easily imitated by general-purpose LLMs. A characteristic of strong Vertical AI companies is that "even if they are imitated, competitors cannot catch up without high industry knowledge and data utilization know-how."
The success of Vertical SaaS combined with the rapid evolution of AI technology has created revolutionary opportunities in areas that were previously difficult to digitize. Industries with language-heavy workflows, particularly law, medicine, finance, and accounting, are already adopting AI, and it is predicted that at least five Vertical AI companies will reach over $100M in ARR within the next two to three years.
The success factors for Vertical AI can be summarized in the following three points:
Access to large-scale markets: Since language-based tasks exist in every industry, the target market is extremely vast
High value proposition: Achieving high ROI by replacing or supporting tasks that are burdensome to perform manually
Differentiation through industry specialization: Demonstrating strong defensibility by accurately capturing regulatory compliance and industry-specific user needs
Players who can identify 'where AI truly creates value' and provide solutions have the potential to grow exponentially over the next few years.
Related Articles
