Latest / Vertical SaaS with Fexingo: Industry-Specific Software for Healthcare, Construction, Legal / How Vertical SaaS Is Going Vertical with AI Assistants
Transcript
- Lucas: So last week I was talking to a friend who runs a construction firm — about thirty people — and he pulls out his phone to show me how his project manager uses Procore. And I notice he's got this little chat bubble icon on the side. I said, 'Is that a support chatbot?' He said, 'No, it's the AI assistant — I just asked it to generate a change order for a client.' Luna: A change order — that's the kind of form that used to take someone twenty minutes to fill out by hand. Lucas: Exactly. And it got me thinking about how the next big wedge in vertical SaaS isn't another feature module — it's the AI assistant. Not a general-purpose chatbot, but a deeply integrated agent that knows the industry's specific workflows, terminology, and compliance rules. Every vertical software company now has a choice: build one, buy one, or get replaced by a competitor that does. Luna: So this is different from the 'AI copilot' buzz we heard in 2023 and 2024, where every SaaS product slapped a ChatGPT-style window on the side. You're talking about an assistant that actually operates on the data inside the system. Lucas: Right. The first wave of embedded AI was essentially a search overlay — you could ask 'What's the status of invoice 402?' and it would pull an answer. Useful, but not transformational. The second wave is agents that take action: drafting documents, updating records, triggering workflows. And the ones that win will be the ones that feel less like a general AI and more like a domain expert. Luna: I've seen this in the legal space. Clio recently launched Clio Duo — an AI assistant that can draft a client intake form, summarize a deposition transcript, even suggest billing adjustments based on the fee agreement. That's not just search. That's an associate-level paralegal built into the software. Lucas: And that's the key insight — these assistants lower the friction for high-frequency but low-complexity tasks. In healthcare, for example, a doctor using Epic might spend fifteen minutes per patient on documentation. An AI assistant that can auto-generate a SOAP note from a voice recording — that's not a nice to have, that's a productivity multiplier that saves hours a day. Luna: But building these assistants requires deep domain knowledge. You can't just fine-tune a large language model on a bunch of medical textbooks and call it a day. The output has to be compliant with HIPAA, with billing codes, with the specific formats that insurers accept. That's a huge moat for companies that already own the data and the workflow. Lucas: That's actually the argument for why incumbent vertical SaaS players might have an advantage over horizontal AI startups. The startup can build a great general model, but they don't have the integration points. They don't know that in construction, a change order needs to reference the original contract clause and the revised budget line. Procore has that context baked in. Luna: So you're saying the winner in vertical AI assistants is likely the incumbent — not a new entrant. Lucas: I think the incumbents have a real shot, but only if they move fast. Look at what happened with Replit. Replit started as a browser-based coding IDE, then they built an AI assistant called Ghostwriter that can generate code from a prompt. They expanded into vertical use cases — for example, helping finance teams write Python scripts for data analysis — and now they're essentially a platform for creating mini vertical SaaS apps on the fly. Luna: Right, Replit is interesting because it's horizontal tech that's being pulled into vertical use cases by users. But that feels like the opposite direction — a general tool that gets specialized by the community, versus a vertical tool that adds AI. Lucas: Both directions are happening simultaneously. And the market will probably settle on a hybrid model — where the platform provides the AI engine, but the domain-specific fine-tuning and workflow integration happen inside the vertical software. The horizontal AI companies like OpenAI and Anthropic become the infrastructure, and vertical SaaS companies become the interface. Luna: That makes sense. But it also means the vertical SaaS companies need to have in-house AI talent, or at least a strong partnership strategy. Not every company is going to build their own model from scratch. Lucas: And that's where we're seeing interesting moves. For example, in the restaurant space, Toast — the point of sale and operations platform — launched an AI assistant that can answer employee scheduling questions, suggest menu items based on inventory, and even generate marketing copy for social media. That's three different workflows, all within one assistant, all tied to the restaurant's data. Luna: And the employee doesn't have to learn a new interface. They just type or speak a natural language request. Lucas: Exactly. That's the user experience promise. But here's the thing — the economics of these assistants are still being figured out. Most companies are charging a flat monthly fee for access, or bundling it into the premium tier. But as usage scales, the cost of compute — the calls to the underlying AI model — could become significant. Some companies are already experimenting with consumption-based pricing: you get a certain number of 'AI credits' per month, and then pay per task beyond that. Luna: That sounds reasonable. But it also introduces a new churn risk. If the assistant is good, users will rely on it. If you then raise prices or cap usage, you might push them to look for alternatives. Lucas: It's a delicate balance. And it ties into the bigger question of how vertical SaaS companies monetize AI without squeezing their own margins. But I think the companies that figure it out will have a massive advantage. Because once a user has trained their workflow around an AI assistant — once the assistant knows their clients, their projects, their preferences — switching costs become enormous. Luna: That's a powerful lock-in. But it also means the assistant has to be genuinely useful. If it's just a gimmick, users will ignore it, and the company wasted development dollars. Lucas: Right. And that's why we're seeing some vertical SaaS companies acquire AI startups rather than build internally. They're buying the talent and the technology, and then integrating it deeply into their existing product. It's a faster path to a genuinely useful assistant. Luna: Do you have a specific example of that? Lucas: Sure. In early 2025, a major healthcare software company acquired a small AI firm that specialized in medical coding automation. That firm's model could read clinical notes and generate the correct ICD-10 billing codes with over ninety percent accuracy. Within six months, the acquirer embedded that into their EHR platform. Now, when a physician finishes a note, the system suggests the codes automatically — no extra clicks. Luna: And that's a workflow that affects not just the physician, but the billing department, the insurance claim process, the revenue cycle. It ripples through the entire operation. Lucas: Exactly. And that's the kind of impact we're going to see more of. Over the next two to three years, I think every vertical SaaS product of any scale will have an AI assistant. The question is not whether, but how good it is, and how well it integrates with the user's daily work. Luna: I wonder if this could also open up new markets. For example, smaller law firms or dental practices that couldn't afford a full-time assistant might now get a software version for a fraction of the cost. Lucas: Absolutely. That's a democratizing effect. And it ties into something we talk about a lot on this show — the idea that vertical SaaS isn't just about software, it's about enabling better outcomes for entire industries. AI assistants are the latest tool in that mission. Luna: Before we go further, I want to mention something quickly. We talk a lot about how these software tools make businesses more efficient, and that's exactly why we do this show — to help you understand what's actually working out there. We deliberately don't run ads on these episodes because we want the content to be clean and focused. If you find these conversations useful for what you're building or running, and you want to support that choice, there's a link — buy me a coffee dot com slash fexingo. No pressure, just a simple way to keep the show ad-free. Lucas: Yeah, and it really does help. We've had listeners say it makes a difference. So if you're inclined, that's the place. Now, back to AI assistants — I want to talk about a risk I haven't heard discussed much: what happens when the assistant makes a mistake? Luna: That's a huge concern, especially in regulated industries. If a healthcare AI assistant suggests the wrong billing code, that could lead to an audit or a denied claim. If a legal assistant drafts a clause that's not legally sound, that could create liability. Lucas: Right. And the vertical SaaS companies are going to have to figure out the liability framework. Some are positioning the assistant as a 'suggestion tool' with a human-in-the-loop — the AI proposes, the human approves. Others are taking on more responsibility through warranties or indemnification. But the legal landscape is still evolving. Luna: I'd imagine that's one reason some incumbents are moving cautiously. They don't want to be the test case for AI malpractice. Lucas: Exactly. But the ones that move too cautiously risk losing to startups that are willing to take that risk. We're already seeing new vertical SaaS companies founded specifically around an ai first approach — companies that don't have a legacy product to protect, so they can build the assistant as the core interface from day one. Luna: That's a classic innovator's dilemma scenario. The incumbent has the data and the customers, but the startup has the freedom to redesign the workflow around AI. Lucas: And the startup doesn't have to worry about cannibalizing a legacy subscription. So they can price aggressively, sometimes even offering the AI assistant for free to build adoption. The incumbents have to decide: do we protect our existing revenue, or do we embrace the new model and potentially disrupt ourselves? Luna: It feels like we're at a tipping point. Over the next year, I think we'll see at least one major vertical SaaS company make a bold move — maybe acquire an AI startup and completely redesign their interface around it. Lucas: I agree. And my bet is that the winners will be the ones that don't just add an AI chat window, but rethink the entire user experience. Think about it: if the assistant can anticipate what you need, do you even need a dashboard anymore? Do you need menus and forms? The assistant becomes the interface. Luna: That's a pretty radical shift. It would change how we think about software usability entirely. Lucas: It would. And it might also change how we think about pricing — from per-seat licenses to per-action or per-outcome models. But that's a topic for another episode. For now, I think the main takeaway is: if you're running a vertical SaaS company, or investing in one, pay attention to the AI assistant strategy. It might be the single most important decision you make in the next eighteen months. Luna: And if you're a user of vertical software, get ready to start talking to your software. It's going to become a whole lot smarter. Lucas: Yeah, it's an exciting time. Thanks for listening, and we'll be back next time with another angle on the world of vertical SaaS.