The Role of AI in Redefining Vertical Software

I recently had an enlightening experience with ChatGPT. In the past, I attempted to learn to code through online courses, only to fail miserably. However, with ChatGPT, I could construct a functional first-draft investment memorandum generator in Python. This experience made me understand the profound impacts that continuous AI advancements will have on every industry. Equally importantly, it provided a deeper understanding of the benefits of vertically focused solutions from the perspectives of users, founders, and investors.

TL;DR

• Vertical SaaS solutions are becoming AI-native. As these vertical solutions gather unique, interconnected data through their various products, they create a powerful “data moat”. This advantage is significantly amplified when integrated with industry-specific systems and databases. 🏰
• Vertical players can amplify the value of their products by utilizing interconnected data across their suite of offerings, creating a network effect where each additional product enhances the overall ecosystem’s performance. 🕸️
• Key elements such as deep customer understanding, tailored UI to customer workflows, targeted go-to-market strategies, and effective communication will continue to play a pivotal role. Each of these elements, on its own, may not provide a sustainable competitive edge. However, their combination, along with the unique data advantage that vertical players possess, can create a robust competitive advantage 🦅
• The combination of AI, data, and industry-specific systems can open the window to create innovative and previously unimaginable workflows🚀

Unleashing the Power of Vertical Solutions

Using the venture capital industry as an example, it’s exciting to think about how AI will enable us to develop tools that can massively increase productivity. This surge in productivity, for instance, could allow VCs to meet more incredible founders in the same amount of time or to close rounds faster without sacrificing diligence quality. Some early experiments with AI VC tooling are discussed in greater detail in these articles (here and here). Other possible/future use cases, not mentioned in these articles, involve leveraging the contextual understanding of LLMs to abstract metrics from portfolio updates and auto-generate KPI dashboards (e.g., alkymi.io), extract data from note-taking tools and pitch decks to generate metrics/valuation benchmarks, analyze data/financial models and generate what-if analysis, assess founder skills from transcripts, and automatically generate LP updates or chatbots to answer commonly asked questions from LPs, founders, or other VCs.

The potential of AI is not confined to the VC industry. Its influence is rippling across all sectors, each exhibiting unique characteristics and product opportunities. My experiences with AI in the VC industry and regular interactions with founders have significantly reinforced my conviction in the bright future of vertical SaaS solutions in the AI era. As articulated by Lightspeed in their SaaS 4.0 article, SaaS is destined to evolve into AI-native solutions over time, with considerable implications for the future of the space.

Building a Robust Data Moat: Customization and Interconnected Data

In an era where data and its contextualization provide a massive competitive advantage, vertically specialized software solutions are well-positioned to reap substantial benefits. To illustrate this, let’s consider an example within the venture capital (VC) industry. Imagine a vertical software startup offering a suite of products. One is a memorandum generator, another is a tool for analyzing financial and operational models and metrics, and a third utilizes past data to generate benchmarks. Each of these products has its unique value, but the real magic happens when they begin to work together.

The memorandum generator, for instance, employs optical character recognition (OCR) to extract data from pitch decks and uses LLMs to generate the memo, potentially saving an analyst up to half their time, depending on the data volume and the prompt quality. However, if this tool could incorporate inputs from the other two tools, its output would become far more valuable. The financial model analyzer could offer nuanced insights into a company’s financial health, while the benchmark tool could provide industry-specific comparative data. The outcome of the memorandum generator thus becomes immensely more valuable with these additional data points. Moreover, integration with industry-specific systems such as Pitchbook, TIKR, Capital IQ, Crunchbase, and Affinity, could further enhance the quality of the output and allow it to provide, for example, advanced exit scenario/probability analysis and in-depth investment recommendations and risks.

In this product ecosystem and with today’s technology stage, while the AI model itself does not inherently ‘learn’ or ‘remember’ data in the human sense, the information it generates during this process can be efficiently stored and retrieved from the vector database for future use. This process creates a form of ‘long-term memory’ for the AI system with the in-context learning approach (as there are more breakthroughs, pre-training may become easier/more common). The system can leverage past data stored, providing a competitive advantage that would be challenging for a new horizontal player to replicate.

Fusing diverse data types and sources generates a comprehensive picture of a business or market, offering substantial value for vertical players. Unique and industry-specific company data reveal the true depth of this “data moat”. Furthermore, they provide a competitive edge in developing, automating, and even enabling new/innovative, value-added workflows that before were not possible (we are still early in leveraging AI in different products and don’t yet completely comprehend all the new flows they will enable beyond just making a current flow better).

In an era where personalization and quality relate to the data, having a company that owns the data across different products that speak to each other allows for better-contextualized model processing that can deliver higher-quality results to the whole ecosystem, making the solutions more defensible and valuable to the customers, creating a product like NFX were the quality of one product gets exponentially better by the addition of another complementary product. As Jerry Chen notes in his Moats article, through a focus on a specific industry, the Vertical SaaS companies of the future will be able to build not the system of record, but the system of intelligence, which will be a valuable position in the market.

If the future ends up being one where there are hundreds of models, a significant additional advantage that vertical solutions may offer is the potential to employ various models (including their own trained ones) and apply the most suitable one to different use cases. As vertical players have a better understanding of the particular uses of their clients than horizontal players, they can choose the most appropriate model for each use case. If this multi-model world does not play out and there are just a few massive model winners, vertical players can still benefit from the same principles of owning unique interconnected private data sets, especially if, on top, they become systems of record in the process.

The personalization component will also become a key area of differentiation as it dives deeper than broad industry knowledge. Imagine a tool that, with the unique data it has, not just understand the overall thought process of venture capitalist but also intimately understands the unique thought processes of individual VCs. These solutions could potentially identify the subtle nuances that differentiate one VC’s investment strategy from another’s, adapting to each unique approach. This level of personalization that having unique data allows drives a higher quality experience and switching costs.

The Role of Traditional Business Fundamentals in an AI-Driven World

Another advantage for vertical plays is that they enable the development of a tailored, user-focused UI, driving a better overall user experience. This allows the creation of a product with a specific user in mind rather than a broad, horizontal solution. It could be a UI that understands the workflow of a particular industry, like a VC’s usual investment process, aligning with their practices and enhancing their productivity. Although a user interface can be easily replicated, creating a UI that truly addresses the unique needs of a particular vertical, respecting its inherent workflow, adds to the overall value proposition of these solutions. However, it’s worth noting that this aspect, while valuable, is not the ultimate differentiator. It is a beneficial add-on contributing to a stronger value proposition when combined with the other mentioned points.

Despite AI’s transformative potential, traditional business fundamentals such as a robust go-to-market strategy and deep customer understanding remain vital. In an era where curiosity and apprehension about AI coexist, vertical players can effectively address customer concerns in their language and speak to their unique pains and fears. AI does not alter the need for strategic customer engagement and effective communication, which will continue to distinguish successful players in this evolving landscape.

Both Vertical and horizontal players will have a key role in the future

All this is not to say that in many scenarios, more horizontally aligned companies will still prevail. These are situations where acquiring additional context from other parts of the workflow provides marginal or no extra value, or simple integration with another tool suffices. In these cases, a more general understanding does the job, and the scalability of the horizontal approach allows for greater cost efficiency.

While some might suggest that companies can bundle separate tools together, the current state of LLM technology and tooling makes this a complex and costly process. Today, large enterprises are struggling to deploy their AI strategies. Now, think of a small SMB like a barber shop; this is simply out of the question for them. From a cost, time, and quality perspective, using one of these vertical providers in many cases will be better. It may be the only option for smaller, more traditional companies that want to benefit from AI.

When the vertical and horizontal debate arises, I believe it’s a matter of returning to the first principles of venture capital, as AI doesn’t change this. Vertical solutions will only succeed when they can genuinely deliver exponential, 10x+ value to their users that more horizontal solutions cannot. In the context of AI, I believe this exponential value will be delivered when the extra contextual data allow for better extraction and a more personalized and quality output.

An exciting future ahead for the vertical software industry

Reflecting on my journey with ChatGPT, I am reminded that we are still in the early stages of realizing AI’s full potential. The true value of this technology will emerge from its intersection with deep domain expertise and the relentless pursuit of 10x better solutions. Vertical Software will undergo a massive change where the AI era will highly amplify these companies' advantages.

As we envision the future of vertical SaaS through the lens of AI, it’s crucial to acknowledge the astonishing pace at which artificial intelligence is advancing. The strategies and viewpoints discussed in this article are shaped by the current state of AI. However, the rapid development and evolution of AI technologies could substantially alter these perspectives in the short to medium term. Agents, for example, are expected to improve in the near future and allow for complex, maybe unimagined use cases that can massively benefit these vertical players.

If you are a founder building in the vertical software space or someone who wants to exchange views, reach out to daniel@flybridge.com