Imagine an HR manager looking for the PTO record of an employee. Instead of logging onto the HR dashboard and pulling up the worker’s profile, the manager asks an artificial intelligence (AI) chatbot to get and email the data.
If this is the future of work, would enterprises still need to use the software as a service (SaaS) HR provider?
This is the question surrounding the SaaS industry in the age of AI. SaaS has defined enterprise software for nearly two decades. It was an upgrade to buying boxed software and running it on in-house computers. Instead, many companies pay a monthly fee per worker for the service.
The SaaS revolution made companies such as ADP, Salesforce, Workday and Slack into multibillion-dollar businesses.
But now another revolution is here. AI could bypass the services provided by SaaS companies. Users might no longer have to use a SaaS dashboard like ADP’s to get the information they need.
Still, the answer is not that simple.
“AI bots won’t replace SaaS entirely, but they will disrupt it in meaningful ways,” Martin Balaam, CEO and co-founder of SaaS platform Pimberly, told PYMNTS. “In some cases, they could even eliminate the need for certain SaaS platforms, particularly larger, slower-moving providers weighed down by technical debt and outdated UIs,” or user interfaces.
Balaam pointed out that businesses pay to use a SaaS tool’s dashboard to work with their data. But with AI, users don’t need a dashboard. They just have to ask the AI to show trends, find patterns and even pull in external data sources that the SaaS tool could not do.
Businesses also pay to have the SaaS tool integrate and orchestrate across the company’s data. AI can also do it. Third, SaaS makes it easier for non-technical users to use the software through drag-and-drop workflows. With AI users, can just describe what they want to do.
“The takeaway: SaaS should evolve with AI, not against it,” Balaam said.
SaaS should be designed so it can work with AI and being flexible so the customer can use whichever AI model works best for them. This flexibility will help SaaS companies endure, Balaam said. The bottom line is that SaaS isn’t dead but the way it provides value is changing.
Zohar Bronfman, CEO of predictive SaaS provider Pecan AI, agreed. “SaaS is not going anywhere, but it is evolving,” he told PYMNTS. “Enterprises will continue to need vertical software designed for their unique data structures, yet the real shift is toward much smarter SaaS that is dynamic, adaptive, and requires minimal customization.”
That means specialized SaaS systems for certain industries have deep, domain-specific knowledge baked in, such as compliance rules, data formats and workflows that the AI assistant might not fully replicate. Like Balaam, Bronfman believes that SaaS itself is changing to become more dynamic and adaptive, becoming even easier to use.
Salesforce was one of the first major SaaS players to go all in on artificial intelligence, launching its Einstein platform in 2016 to embed predictive intelligence directly into its CRM. It doubled down in recent years, weaving generative AI into Slack, Tableau and Service Cloud. Workday, ServiceNow, ADP, Slack and other SaaS providers also are putting AI into their products.
See also: Klarna CEO: SaaS Companies to Consolidate as Customers Seek ‘Hub of Knowledge’
Now for Agents as a Service
But Rahul Sharma, CEO at HSBlox, thinks traditional SaaS can be replaced, by agents as a service (AaaS). He told PYMNTS that agents can outperform SaaS systems, which basically let companies create, read, update and delete records through a user-friendly interface.
“AI agents will be able to understand what users want or need, anticipate their requests, and eliminate the need for the current model of SaaS applications,” Sharma said.
Sharma gave this example to illustrate how SaaS is different from AaaS:
Task: Claims processing
SaaS: Identifies issues resulting in denial of claims, but still requires action from healthcare workers to trigger different resolution workflows
Agentic AI: Automatic validation of claims, identification of any missing information, triggers any workflows that require resolution, and reduces denials. AaaS uses LLMs for clinical documents interpretation and extraction and matching for coding accuracy.
However, agentic AI still needs to earn the trust of enterprises. According to a July 2025 PYMNTS Intelligence report, 85% of chief financial officers (CFOs) surveyed have no plans to deploy agentic AI. And even among the 15% who do, 89% remain in the evaluation phase without a concrete roadmap.
Whether or not AI agents are ready for prime time, Jarie Bolander, general manager and executive partner at Decision Counsel, believes SaaS is on its way out. “Traditional SaaS is dead,” Bolander said. “Single endpoint solutions are rapidly going away. When you can build pretty much anything you want with AI, then what matters is the outcomes that companies can provide.”
Seymour Duncker, AI and ML executive strategist at Decision Counsel, added that with SaaS, users still do all the work. With agentic AI, the agent does the work for the user. “Everything will be outcomes or transactions-based. … So yes, traditional SaaS is dead.”
Not all agree. Some see AI as the natural advancement of SaaS. “AI is not the new SaaS. It’s the next generation of it,” Greg Bibeau, CEO of Terminal B, told PYMNTS.
“SaaS revolutionized how businesses consumed technology by making it scalable, available, and subscription-based,” Bibeau said. “AI is now doing for SaaS what SaaS previously did for on-premises software — redefining the value proposition.”
Whether SaaS is “dead” or simply evolving, there is at least one consensus: The subscription model that once defined the software industry is giving way to an outcomes model, where users expect AI to not just provide tools but deliver results.
“The next generation of SaaS will be judged not by how many features it offers, but by how much real work it takes off the customer’s plate,” Mike Cieri, general manager of software solutions at BILL, told PYMNTS. “That means fewer clicks, less effort, and more results.”
Source: https://www.pymnts.com/
