Generative AI is Coming for Vertical SaaS
Generative AI is moving so fast, it’s hard to keep up! This is incredibly powerful technology with a stunning pace of innovation across a complex value chain. But despite the rapid progress, thus far the impact on vertical SaaS, a major area of focus for us at Pender Ventures, has been minimal. In this piece, I share some thoughts on how companies that build vertical SaaS products can think about generative AI and how things might play out going forward. Let’s dig in.
Defining vertical SaaS
Vertical SaaS is software built for a specific industry. You could call it customer-first SaaS, because it takes the needs of a specific set of customers as its starting point. By contrast, horizontal SaaS is software built for a use case or problem that is relevant across a range of customers. For example, the problem of managing customer relationships can be approached horizontally, as in the case of Salesforce or HubSpot, or it can be approached vertically, as in the case of Jane (CRM for allied health) or Spark (CRM for real estate).
The first wave of generative AI is horizontal
The first wave of generative AI companies have been almost exclusively horizontal. Building on top of powerful foundation models, these companies have reimagined core workflows for text and image generation from the ground up. Here is an example of some of the hottest horizontal generative AI companies:
Company | Product | Last Financing Round |
Runway ML | $50M, December 2022 | |
ImagenAI | photo editing + post-production | $30M, December 2022 |
Descript | media editing | $50M, November 2022 |
Mem | note-taking | $23.5M, November 2022 |
Jasper AI | copywriting | $125M, Oct 2022 |
Typeface1 | copywriting | $65M, Feb 2023 |
If you look at generative AI market maps, you will find many similar startups building for broad, horizontal use cases. And if you zoom out to more established horizontal companies like Adobe, Notion, Canva and Slack, you will find them rushing to integrate generative AI into their existing products.
Vertically focused generative AI companies are coming soon
So, will we see generative AI companies with a vertical focus? I believe the answer is yes, and they will be here soon. After all, the customer-first mindset of vertical SaaS remains just as valid in the generative AI era. As Sarah Guo, a leading AI investor, puts it:
There are two reasons why vertically oriented products have not been first out of the gate. The first is that, even if GPT4 can pass the bar exam, foundation models are necessary, but not sufficient, to produce high-performing vertical SaaS. To deliver high quality construction schematics, financial prospectuses or legal briefs—to choose just a few examples —product builders will need to assemble industry-specific data sets that can refine the output of the foundation models. Once assembled, they will need to use Reinforcement Learning from Human Feedback – the same technique that brought ChatGPT to life – to train their apps to be aligned with what users expect, meaning that they provide consistently high-quality results without spooky stuff like hallucinations.
The second reason that vertical products will take longer is for all the reasons noted in Sarah Guo’s tweet. Product, UX, and go-to-market all matter, and the more specific the customer is, the more these issues come to the fore. This is why great product and business teams will be so important in bringing generative AI to vertical SaaS. Foundation models, made accessible via apis, are swiftly becoming part of the new software stack, like databases, cloud and big data before them. Great founders will need to assemble strong teams of applied AI developers to build on top of them, (while avoiding the pitfalls of chasing strong researchers who don’t want to build products), but in vertical SaaS, deep customer insight and go-to-market expertise will remain paramount.
How should existing vertical SaaS companies think about generative AI?
Most existing vertical SaaS solutions were not built with generative AI in mind. These customer-first software products are now in market, serving their users. They have features to maintain and a roadmap to follow. At the same time, these companies have an opportunity to win big in this new technical paradigm, particularly given their distribution advantage. They also need to be careful to avoid being disrupted by new entrants that are taking a fresh technical approach to their space.
Here are a few guiding principles for customer-first vertical SaaS companies navigating this new technology:
1. Internal productivity is the lowest hanging fruit
The internal work of building vertical SaaS companies is an easy place to begin leveraging the power of generative AI without touching the core product. How might you use this technology to summarize meetings and customer calls, create internal documents and presentations, and run internal analytics? Products that move the needle on efficiency are already on the market and worth investigating.
2. Go-to-market is another potential early win
Generative AI is actively reshaping marketing. One of the hottest areas of activity is content marketing – a bellwether of demand generation for vertical SaaS. Email marketing is at the front of the line too, as are ad copy and images. Customer support is another massive opportunity. As with internal productivity, there are already solutions available that could be baked into your existing go-to-market playbook.
3. When it comes to core product, think tech first, customer second
Vertical SaaS companies typically start with the customer, but generative AI is about rethinking the software experience by turning human users from creators into prompters of AI systems and editors of their outputs. So, to take advantage of this new technology, vertical SaaS companies should think about what parts of the software experience might transform based on these capabilities, and then think about how those changes could add value for customers. More on this below.
What use cases are relevant when applying generative AI to vertical SaaS?
Text and image generative are extremely broad categories. For example, even software code falls within the spectrum of writing, which is why products like github copilot are so transformational. With this broad scope in mind, here are a few areas that vertical SaaS builders can dig in on as they think about applying generative AI to their products:
1. Filling forms – A lot of the writing in software is filling out forms. Replacing this often-tedious work for human users will be most welcome and is already quite possible. For example, software companies that help companies with security and compliance are already hopping on the opportunity
2. Summarizing – Large language models are not just good at generating new text, they are also proficient at boiling down substantial amounts of information into succinct and digestible summaries. Whole companies are being formed around this ability, which may be relevant to certain SaaS workflows.
3. Human-computer interactions – Large Language Models excel at understanding human text. This capability enables a new modality whereby users can switch from menus and explicit inputs to natural language. Christoph Janz, a leading vertical SaaS investor, has a great description of what this looks like here.
If you’re looking to delve deeper into these themes, this piece on applying generative AI to fintech contains several interesting perspectives.
What are we looking for at Pender Ventures when it comes to generative AI and vertical SaaS?
If you are a customer-first vertical SaaS company, we are looking for you to have a good answer for how you are approaching generative AI in your internal operations and potentially in your core product. We also want to see that you have thought about how these technologies might change the competitive dynamics in your industry. If you do not see generative AI impacting your core product in the short run, that’s fine, but be careful that you are not thinking too narrowly. If your answer is “we haven’t thought about it at all,” that is probably a red flag.
For the next wave of generative AI-first vertical SaaS, our main questions will be around competitive advantage. One obvious source of advantage will be a unique data set and a set of users that can help train your models on that data. But data moats are just one approach; it might also make sense to argue that you have a durable first mover advantage, a better product/UX, or unique domain insights. In other words, traditional paths to winning will remain valid in the era of generative AI.
Will generative AI have a big impact on vertical SaaS?
Yes, it’s going to have an enormous impact. For existing players, as leading SaaS investor Thomas Tunguz puts it, “ML will become a requisite feature in most workflow SaaS”. Looking further out, net new companies with a generative AI- first approach will leapfrog established players in certain industries. At Pender Ventures, we are looking forward to being a part of this evolution. We are especially interested in companies applying generative AI to health tech, a tricky problem given the level of accuracy needed. So, if you’re working on applying generative AI to vertical SaaS, we would love to chat!