AI & its Impact on Vertical SaaS: The Gates of Distribution have Flung Open

With the latest boom of AI innovation driven by large language models (LLMs), there has been a cambrian explosion of vertical SaaS companies capitalizing on new capabilities enabled by the latest generation of Large Language Models (LLMs).

As someone who has been excited about the next generation of vertical saas for some time, it’s been an incredible time to be part of the startup ecosystem. I’ve met with companies building for a wide variety of use cases from legal to healthcare to financial services to construction and everything in between.

What is driving this boom? What are the new capabilities that LLMs are unlocking?

There has been a lot of hype about AI agents and the ability to automate previously manual workflows. While I believe there is true potential to do aspects of this, I’d argue it is often hard to completely automate away large swaths of human tasks. Instead, I’m equally, if not more so, excited about the potential impact of doing one step before this — to make humans more efficient by speeding up processes and making traditional workflow software more delightful.

These steps before full automation can still have a significant impact in verticalized use cases. They enable humans to be more effective and free them up to focus on higher order work. Breaking this down, I think about AI as having the potential to impact the following sets of tasks:

• Data structuring — The first step required for automation is understanding the underlying data. In order to do this AI tools need to be able to structure the data to make sense of it. This can be in the form of invoices, contracts, referral documents etc. High quality AI is able to take this previously unstructured data and turn it into something it can read and understand systematically with high accuracy.
• Analyzing — Once the data is readable, the next step is to analyze the “what”. What is this data that the AI is ingesting and what does it mean. For example, understanding the clauses in a contract and even being able to point out how certain contracts differ from a more standard template.
• Recommendations — Another downstream action that AI can take after understanding and analyzing data is being able to provide recommendations. For example, it can suggest potential diagnoses based on symptoms a patient shares during a doctor visit or recommend improvements to human written code to make it more effective.
• Taking action — AI agents can now take action on their recommendations. For example, an AI agent could reach out to a supplier to check in on their progress after predicting that the production of a company’s flagship product could be delayed. Or it could call a patient’s healthcare insurance to verify that she is eligible for a procedure on behalf of the clinician.

Importantly, when AI is embedded into a workflow it does not have to do all four tasks to be impactful. Even a subset of these can have a huge effect. For example, imagine the efficiency gains of software being able to analyze a company’s customer agreements and surface any coming up for renewal in the next quarter. Or auto categorizing financial transactions instead of human accountants doing so manually to close a company’s books at the end of every month. Magical!

Unlocking distribution

I’ve previously written about what I believe to be the characteristics that make a great vertical saas company. While (1) a large enough TAM, (2) building critical workflow software and (3) solving a real need are still as important as ever, I believe we are at a unique point in time because the cost of acquiring a new customer is at its lowest point in at least a decade.

Yes AI has made products more delightful and more useful, but it’s often not the only reason for this declining friction in sales. For the companies that are building true integrated workflows (more here about why that is important) AI enabled tools are only part of a broader software workflow that would be adopted to make it core to a user’s daily work.

So if product advancements are not the largest unlock, why is there more opportunity today? I believe that it’s because AI has unlocked distribution — potential users are actively seeking out AI solutions, making this a uniquely promising time to build a vertical SaaS company.

AI has captured the average user’s imagination of what is possible. People, regardless of how tech savvy they are, have bought into the promise of AI and believe it can be transformative to their workflows. From my customer conversations with a broad swath of professionals from lawyers to service technicians to doctors, they are more open to new software than I’ve ever experienced.

Take lawyers for example. These buyers are notoriously challenging customers to acquire. They are not usually early adopters of technology, often priding themselves on being “late”, after the kinks and risks of a potential solution have been worked out. However in today’s market I’ve seen how quickly potential buyers are moving to schedule demos and launch pilots to implement new AI products. This speed and willingness to test emerging solutions illustrates just how drastically sentiment has changed even in verticals that were previously extremely difficult to penetrate.

Defensibility

It is still early days for the vast majority of AI forward vertical saas companies that have emerged over the past 18 months. The AI that most are embedding in their solutions is not what is driving long term defensibility — it is arguably largely “off the shelf” technology with a bit of customization around the edges. The combination of successfully building distribution early and building truly embedded workflows for their users is what will lead to enduring businesses long term.

In addition the proprietary, vertical specific data that a company accumulates over time will allow it to improve a generic AI solution to be even more effective for its specific use case. For example, while a horizontal LLM may be very good at structuring invoice and appointment data for an HVAC vendor, it cannot analyze which types of issues are most common by heating system and which technician is best to service each without the proprietary information amassed by working with companies in the vertical. Therefore if a company can scale by earning the right to be the core software solution for its customers, the data it accumulates over time will enable a defensibility flywheel — improvements in its AI capabilities, which in turn will drive a more meaningful product moat in the medium term that should drive further distribution and lock in.

As a venture investor that spends all of my time with businesses early in their lifecycle, I’m seeing a renaissance of companies going after vertical markets. Founders are feeling market pull from users hungry for AI solutions to make their day to day lives easier. There’s never been a more clear “why now” because the gates of distribution have flung open like never before.

If you’re building verticalized core workflow software in any industry, reach out. I’d love to connect!