News / SaaS pronounced 'dead' as AI moves in – but ROI is the trick

ID 427518234 © BiancoBlue | Dreamstime.com

“SaaS is as we know it is dead”, Manifest 2026 attendees last week were told – the version of the model built on standardised workflows, static user interfaces, and one-size-fits-all roadmaps is becoming obsolete.  

The shared message from The AI Pivot panel, comprising Pablo Palafox, CEO of Happy Robot; Mathew Elenjickal, CEO of FourKites; and Krener Komoni. CEO of Tive; was that enterprise software is shifting from selling tools to delivering outcomes – with AI agents increasingly doing the work.

And this shift is forcing vendors to rethink product design, implementation, and pricing. 

“When you think about SaaS, we are optimising for standardisation… But in the AI world, we are really optimising for outcomes,” said Mr Elenjickal, adding that outcome-driven automation requires systems to be “nimble”, and configurable in ways traditional SaaS platforms were never built to support. 

“We have to be able to configure the systems and customise the systems. It’s very different than the SaaS world. So yeah, absolutely, we have to pivot quite a bit,” he said.  

The sharpest claim came from Mr Palafox, who stated: “SaaS, as we know it, is dead.”

“It’s really because you’re going to be able to personalise your software, especially at the enterprise level, downstream as well…. Not only a personalised workforce, but also the instrumentation and the dashboards that agents are going to be building to show the work they are doing, that’s going to be personalised as well,” he explained. 

“Just because we’re doing things in a certain way doesn’t mean we have to do them the same way now. Meaning, just because you were buying some software that is static among all of the customer base of a certain vendor, doesn’t mean that today you have to agree to that or accept that fact.”

Mr Palafox continued: “Customers don’t think about it as workflows anymore.  

“They think about it holistically, and they don’t have to really think about the details, because those details, agents themselves are going to fix.  You don’t go telling your humans all the minutiae they have to do every day. You shouldn’t. They should be self-guided. Well, AI is going to abstract that away. It’s already abstracting that away. Agents are improving themselves.” 

Happy Robot provides agentic infrastructure for an enterprise AI to build its own workforce, and despite having launched just four years ago, Mr Palafox underscored the pressure of keeping up with today’s standard.  

“Every hour there’s something that changed. and you have to reinvent yourself,” he said. 

Mr Elenjickal highlighted a specific pivot in order to recoup return on investment into agentic AI. He explained: “Our entry into the agentic world started 18 months ago by automating various specific workflows for customers. And obviously, the ROI from AI comes down to either taking cost out, people out, or helping companies scale by adding those people.

“What became very clear is that automating just one workflow is not enough to get an ROI.

“Because, a human being is doing 10, 15, 20 different things and, unless you are able to automate everything, or the majority of what they do, there is no clear ROI,” he said.  

One thing Fourkites had to do within its platform was “ingest any process and build automation around it” – not just processes within Fourkites, but “any different systems, any different teams, any external partners”.  

He added: “That flexibility had to be built in to really deliver ROI, because customers are like, ‘yeah, you can automate one thing; that’s a great science experiment. But it’s not a real thing’.” 

Mr Elenjickal added: “AI infrastructure where you can get ROI; that is the pivot we had to make.” 

This article is © The Loadstar. Reproduction, rewriting, or derivative use requires a license. Contact [email protected] for licensing enquiries.