AI Is Coming for SaaS. Just Not the Way You Think.

A B2B SaaS CEO’s Perspective on the SaaSpocalypse

The question every SaaS CEO is getting right now from their customers is: “Why buy your software if I can just try to build it myself with AI?” The question every SaaS CEO is getting from their Board or investors is: “What are you doing to futureproof the value of your proposition?”

These both point to the now daily evolution of what’s possible and the potential for disruption in SaaS businesses. The inquiry deserves a straight answer.

Yes, AI is going to disrupt some SaaS businesses.
Yes, the value of code is collapsing.
Yes, some categories are likely to get hit hard.

If your entire value proposition rests on “we built the product for the last XX years,” you should be nervous; developing the product is no longer the hard part.

We’re seeing it in conversations we’re having with buyers everyday. A year ago, most conversations started with “What’s your AI strategy?”, eager to take advantage of whatever was on offer and invest with little promise of ROI. Now they’re asking something more pointed. Are they locking themselves into something that could look obsolete in a few years? Could they build something close enough internally, faster and cheaper? Why should they commit while the platform war rages on?

Here are some dimensions of evaluation to answer the pressing questions above.

First: Cost vs Value

Every SaaS business operates with a gap between what customers pay and the value they can measure from using the product. That “value gap” is what allows you to invest, grow, and sustain the business over time. The wider the gap, the more resilient the offering.

AI dramatically compresses that gap. And products that have gotten progressively more expensive over the years without a paradigm shift in value will be exposed.

Which means every SaaS company should have an imperative to continue to compress their cost of delivery to maintain this gap. History indicates that “good enough” and dramatically cheaper often prevails, when the cost of building drops as quickly as it has, “good enough” becomes a very real alternative.

Second: Horizontal SaaS vs Vertical SaaS

The most vulnerable SaaS companies are the ones that are broad, shallow, and easy to approximate, Horizontal tools built to solve a simple use case at scale across nearly every market, without requiring a deep understanding of how their customers actually operate.

That flexibility has historically been a strength. But in a world where software can be generated faster and cheaper, it becomes a vulnerability. If your product can be approximated with a set of configurable workflows and basic logic, it is only a matter of time before your customers, or someone else, recreates something that meets their needs at a fraction of the cost.

That puts expensive, horizontal SaaS platforms directly in the line of fire. The broader the use case, the easier it is to replicate. The shallower the domain expertise, the less there is to protect. If your differentiation is flexibility without depth, AI is not your friend in the near term.
This is where a lot of the current conversation misses the point.

Which is why the real question isn’t whether AI can build software. It can. That’s already clear. The better question is whether it can build software that actually understands how a business runs.

Because the value of a SaaS company is not the code. That’s the piece getting commoditized. The value is the business process logic. The accumulated understanding of how a specific type of company operates, encoded into workflows, data models, and increasingly, into the context layer that AI depends on to be useful.

That’s what vertical SaaS actually is. And it’s why it holds up differently under this kind of pressure.

Because, with vertical SaaS, you’re not just buying software. You’re buying a point of view on how your business should run, shaped by patterns across hundreds of similar companies, refined over time, and embedded into the system itself. That’s not something you replicate quickly, no matter how good the coding tools get.

Third: Innovation vs Automation

There’s another category of risk beyond horizontal platforms, and this category is based not on what the companies in it sell, but around how they respond to this moment. A lot of companies think they have an AI strategy, but it’s a head fake. What they have is automation of the status quo.

There’s a lot of excitement around agents right now. And, yes, you can absolutely build an agent that performs a task faster than your software vendor can ship it. That’s real. One of our largest customers said it plainly to our Chief Product Strategy Officer recently: we just need to build it for our use case, you have to build it for everyone.

But speed isn’t the constraint. Context is. As Kantata’s CTO Vikas Nehru put it recently:

“If you give an AI an “Agent,” you’ve given it a car. If you give an AI a ‘Knowledge Graph,’ you’ve given it a map. Currently, we are all collectively cheering for a bunch of cars driving 100mph in total darkness.”

The currency of the future is context. AI without context isn’t nearly as valuable as people think it is, and context is not something you spin up overnight.

Which is why a lot of what’s being delivered right now is speed without understanding. Product roadmaps are being filled with agents that automate work that already exists, instead of making the system itself more intelligent. The distinction is whether the software is actually helping the customer transform their business model, or just doing the same things faster.

I see this manifesting every day in the professional services sector. These businesses are defined by a constant cadence of projects with complex revenue and client dimensions. Revenue depends on how work is staffed, how it’s billed, when it can be billed, when it can be recognized, and how forecasting holds up when reality hits. These are not independent workflows. They are tightly connected, and small changes in one part of the system ripple across everything else.

It is entirely possible to automate pieces of that. In many cases, that is the right move.

But building a system that actually reflects how that business operates, and can support decisions across that complexity, requires context.

Gartner® put a finer point on this in its recent research on domain-specific language models. As their research notes, “the use of your proprietary context produces results that won’t be replicated by a generic LLM.” (Gartner, DSLMs Are the Future of Service Delivery Intelligence, by Danny Ryan, 18 February 2026)

But Gartner also mentions that “Not every technology services leader has the internal resources or time to build a DSLM from scratch.” We believe for many organizations, especially those without deep AI resources, leveraging vendor-delivered domain models embedded in systems like PSA is the faster and more practical path.

Kantata’s Expertise Engine is built on this exact principle. Instead of asking every company to build and maintain their own domain model from scratch, we embed that capability directly into the platform, using delivery patterns and real operational data to drive better outcomes without requiring a full internal AI program.

Even now, building the kind of domain-specific AI systems that make this work at scale is out of reach for most companies. Some of the largest enterprises will try. Most companies won’t have the resources, the data discipline, or the time to get there.

And even if they do, maintaining and evolving those systems becomes a permanent responsibility. The cost of software is only one part of the equation. Implementation, change management, time to value, operational risk – these don’t go away because you wrote some code. If anything, they get more complicated when you own the entire stack yourself.

Which is supported by the irony that the companies at the forefront of Generative AI still run on SaaS systems internally. They’re not abandoning software. They’re integrating AI into it, because systems of record that reflect real business workflows still matter. These models depend on systems of record. Without that foundation, none of this works.

So where will this leave SaaS? Not dead. Not untouched either.

The easy parts of software are getting easier. The hard parts are becoming more visible. Pricing models will evolve. Expectations around speed will rise. The line between what a customer builds and what a vendor provides will blur.

But the core requirement does not change. Businesses need systems that understand how they operate and can translate that into decisions, workflows, and outcomes. AI amplifies that need. It doesn’t remove it.

If you’re a SaaS CEO, this is not a moment to downplay the change. It’s also not a moment to panic. It’s a moment to be clear about where your value actually sits.

If it’s in the code, that’s a problem.

If it’s in the context, you’re building something much harder to replace.

GARTNER is a registered trademark and service mark of Gartner, Inc. and/or its affiliates in the U.S. and internationally and is used herein with permission. All rights reserved.

Gartner does not endorse any company, vendor, product or service depicted in its publications, and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner publications consist of the opinions of Gartner’s business and technology insights organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this publication, including any warranties of merchantability or fitness for a particular purpose.