AI Won’t Kill SaaS—It Will Make It Smarter
Lately, there’s been a lot of noise about how AI will make the current SaaS model obsolete. Satya Nadella has gone so far as to say that the very concept of business applications will "collapse" as AI agents take over. These are bold claims and one needs to examine them critically to separate the hype from reality.
The Hype vs. Reality
There seem to be two core ideas behind the predictions of SaaS’s demise, and these ideas might sound compelling until you consider the realities on the ground:
HYPE: Building custom software tailored to a business’s specific needs is going to be so inexpensive that firms will stop using SaaS platforms. After all, why go for an expensive generic solution if you can just build one cheaply by business users speaking to an LLM powered coding assistant?
REALITY CHECK: One can easily be misled by LLMs outperforming top human coders at many benchmarks, or by flashy demos where LLMs seem generate an entire working app from a simple prompt. High scores on standard coding benchmarks haven’t translated to excellence in an enterprise software development environment. You would still need skilled developers that can use these LLMs as coding assistants, and coding is just a small part of the overall cost that goes into building and maintaining an enterprise business application. Someone still needs to define the requirements, create the functional specs, write and execute the test cases, and maintain the underlying infrastructure.
REALITY CHECK: Now, there are of course some scenarios where the SaaS offering is so straightforward (e.g., it does a simple but useful task for you and charges, let’s say, $5/month) that it can very easily be replaced by a custom AI developed solution. But this isn’t the space where the bulk of the major SaaS providers operate today.
HYPE:Autonomous AI agents powered by large language models (LLMs) will replace traditional SaaS platforms. Why click through a CRM when an AI agent can manage contacts, pull reports, send emails, and maybe even respond to emails for you?
REALITY CHECK: Satya Nadella called SaaS applications as “essentially CRUD databases with a bunch of business logic.” He then follows on to say “the business logic is all going to agents,” and of course all that’s left of SaaS is a “bunch of multi repo CRUD.” These statements massively understate the size and complexity of business logic involved in many SaaS solutions. Defining and maintaining all the business logic, workflows, and then maintaining them in an ever-changing business, compliance and regulatory environment will remain an expensive proposition. And many enterprises will still find the SaaS model cost effective as they are essentially sharing the cost of development and maintenance with many other enterprises.
REALITY CHECK: Even if some agents can be built to contain all this business logic who’s better placed at this than an existing SaaS provider? And of course, the more likely scenario is that most AI agents become a layer on top of SaaS, orchestrating actions across multiple platforms.
Conclusion
While AI will undoubtedly change the way we work with software and how software systems interact with each other, I don’t buy into the idea that SaaS is doomed. People underestimate the complexity of many SaaS applications, and overestimate the capability of LLMs to build and maintain these applications in-house cheaply and reliably. Instead, I see a future where SaaS and AI co-evolve, making business software smarter and more agentic rather than SaaS disappearing entirely. We will still have SaaS providers, but the solutions they provide on will have AI integrated throughout the application.
The most realistic future isn’t one where AI makes SaaS obsolete, but one where SaaS and AI blend seamlessly. What’s actually happening is that AI is becoming a native part of SaaS platforms. Microsoft, Salesforce, and other leaders are integrating AI assistants directly into their software, making them more intelligent and agentic. The SaaS model will evolve, incorporating more AI-powered automation, personalized insights, and intelligent interfaces. Some workflows may shift away from traditional UI-driven apps to conversational AI interactions, but the backbone of business software—business logic, security, compliance, and structured data—will still live within SaaS platforms.
Like this article as it is very relevant to my field. My thinking is that SaaS will remain great option for mid size companies. Some of the large companies may go for custom ( like before mid 90s)as their requirements are very unique and security needs are very tight.