KEY TAKEAWAYS:

• SaaS revolutionized enterprise software with easy deployment and scalability, but its inefficiencies—rising shelfware, data silos, and per-seat pricing—are now major pain points for businesses.
• Agentic AI solves for SaaS’ core limitations, eliminating data silos and automating workflows end-to-end, making software more intelligent, autonomous and outcome-driven, rather than just assisting humans.
• Pricing models will evolve as Agentic AI adoption increases, with SaaS’ seat-based model being replaced by usage- or outcome-based pricing, resulting in significant cost savings and improved RoI for enterprises.

In 2011, an essay called “Why Software is Eating the World” captured the public imagination. In it, venture capitalist and inventor of the first web browser, Marc Andressen, argued that software would ‘eat’ into every part of the economy, transforming infra-heavy industries like retail, telecom, and entertainment, with ease of access and instant delivery.

For the enterprise world, that transformation was led by SaaS, or Software as a Service. SaaS moved software to the cloud, becoming a revolutionary alternative to cumbersome on-premise solutions. It did away with lengthy installations and the need for ongoing maintenance, enabling ‘one-click’ deployments, on-cloud collaboration, and scalability.

But fast forward to 2024, and what we’re seeing is more complicated. Businesses are grappling with bloated SaaS systems, with a 2022 survey reporting that 42% of IT professionals believe their most crucial challenge is finding unused or underutilized SaaS licenses within their business. About a third believed that 20-39% of their SaaS spending was wasted.

How did we get here? And as SaaS companies rush to integrate GenAI into their solutions, what does the future of enterprise software look like?

The landscape is shifting again. But instead of more co-pilots and chatbots within SaaS, the future lies in Agentic AI. This new technology stretches what we think software is capable of, by actually “doing the work”—executing workflows end-to-end, optimizing tasks, and delivering business outcomes—instead of just being tools that aid (and sometimes complicate) human work.

And as this new paradigm emerges, traditional SaaS stands to become obsolete in comparison to outcome-driven, autonomous AI agents.

The Rise and Fall of SaaS

There’s no doubt that SaaS revolutionized enterprise productivity: It liberated IT teams from tedious on-prem installations, security patches, and complex maintenance work. Enterprises could now deploy and scale their software easily, by enabling access for each ‘seat’ or user who needed it, with asynchronous collaboration between teams and functions via the cloud. As of 2023, SaaS was valued at USD$273 billion, with businesses running an average of 371 different SaaS applications.

This market is estimated to grow to USD$1.32 trillion by 2032. SaaS has gradually become the third largest cost center for many companies, right after staff and office costs. A significant proportion of these purchases fall under shadow IT costs (i.e. outside official IT governance processes). The very features that made SaaS successful, like on-demand installation and per-seat scalability, have resulted in a “SaaS sprawl”, with businesses struggling to map their SaaS spends and the RoI from it.

Now, especially after the overspend on software through Covid-19, businesses are under pressure to downsize and ensure tangible RoI, with some companies cutting software budgets by 30%. There is real fatigue—the average cost per employee on SaaS tools is $3500 per year—causing internal scrutiny on underused seats and licenses. Where there should be automation, there now seems to be excess.

The other big problem with the current state of enterprise SaaS is that of data silos. Businesses rely on SaaS applications to manage core processes, but many of those applications don’t integrate with each other. For example, marketing and sales may function on overlapping sets of data, but because they are stored in separate applications, this results in complications tracking down the right data, dealing with duplicate or inconsistent records, and a loss of efficiency. The cost of this coordination overhead, according to a 2022 Forrester study, is that knowledge workers are spending an average of 12 hours per week “chasing data”, which could be better spent on more value-added tasks.

Furthermore, because SaaS apps exist in fragmented environments, the organization is unable to make use of all its intelligence on any given customer, across departments and data sets, and accordingly act on relevant information. And the addition of GenAI co-pilots and chatbots within SaaS won’t necessarily solve these issues, because the data silos will still exist, and human employees will have to coordinate workflows across them.

SaaS sprawls, but the fundamental problem remains: that of actually simplifying work in order to deliver tangible business outcomes.

The Next Gen of Enterprise Software: Agentic AI

Agentic AI addresses these core limitations of traditional SaaS. Autonomous AI agents can execute complex workflows end-to-end, and they’re capable of reasoning, planning, iterating on their own outputs, and getting better with feedback and time — like humans.

Unlike fragmented SaaS tools that require manual coordination and exist in silos, Agentic AI can orchestrate tasks across a function to deliver actual output. Let’s take a common HR workflow, for example. With traditional SaaS, an employee onboarding would involve multiple steps and tools: enabling access to the right systems and accounts, another for sending mandatory emails and documents, a third for initiating payroll, etcetera. These tools exist in silos, and HR teams manually coordinate tasks and data across them.

But with Agentic AI, the process becomes more seamless. An agentic HR system would automatically retrieve candidate information, set up employee accounts across systems (communication, payroll, benefits), and guide the new employee through each step of the onboarding process, without manual intervention. It might then check in on the employee periodically, track their performance over time, and also learn to interact with other agents handling other aspects of the HR function.

By eliminating the need for multiple separate SaaS tools which are handled manually today, Agentic AI systems can “do” workflows across the organization holistically.

But here’s the even more transformative implication of Agentic AI: there’s no need for the traditional seat-based pricing anymore. SaaS grew because of a pricing structure where companies paid for each user or license to access the software, rather than a lump sum amount as was prevalent pre-SaaS. While it seemed like SaaS had solved the software pricing problem by moving to a more usage-based model, it turned out that most users don’t actually use the software with the same intensity. So enterprises are stuck with a massive number of seat-based licenses, many of which are rarely used.

But with Agentic AI, enterprises can start paying for what the software actually delivers. For example, in the HR onboarding case, instead of paying for multiple SaaS tools, you may only be charged for onboarding workflows actually executed, or the outcome delivered, such as the number of employees onboarded. This can drastically reduce wasted enterprise costs on SaaS. And it would align software expenditure very closely with actual business value.

With such usage- or outcome-based pricing, companies don’t have to worry about paying for idle licenses across multiple SaaS apps. Instead, they can pay for the work that agentic AI systems complete—whether it’s onboarding employees, resolving a certain number of tickets, or generating and closing sales leads, or all of them.

Agentic AI is thus going to usher in a new business model in enterprise software, which will ensure that businesses get RoI from their software investments, as the cost is tied directly to the AI’s contribution to measurable business goals.

The Challenge: SaaS’ Innovator’s Dilemma

Now, this raises a fascinating line of inquiry: What kind of outcomes should be tracked, and how should they be priced? How will enterprise software be developed and sold in the new landscape? How should the modern enterprise evolve to unleash the full potential of Agentic AI?

And while answers to these will follow, one thing is for certain for now: Traditional SaaS is on its way out. In his bestselling 1997 book called The Innovator’s Dilemma, Harvard professor Clayton Christensen wrote about why many well-established companies struggle to embrace disruptive innovations. Incumbents are so focussed on maintaining existing revenue streams, derived from their core services and business model, that they fail to adapt to new technologies and business models — because it would cannibalize their current profits.

And in the case of SaaS companies, the dilemma is sharp: Shifting to Agentic AI would move SaaS away from per-seat pricing, a model that has been incredibly lucrative for them for decades. Already, we’re seeing companies adopt some form of usage-based pricing, or a hybrid of per-seat and usage- or outcome-based pricing, in order to not be left behind. But even here, the AI on offer is often an “add-on”, or for a vertical use-case.

We at Ema imagine a very different future for the enterprise stack, in which multiple autonomous AI agents flourish as an integrated part of the enterprise, working together intelligently, like the organization’s horizontal operating system. This helps cut down the software sprawl and data silos we’ve seen in the current SaaS era. In this new enterprise of the future, outcome-based pricing will be aligned with actual business needs. And for traditional SaaS, the transition isn’t just about adding new technological capabilities. It is also about overhauling the core business model and strategy, which attacks current revenue. SaaS will now be forced to charge for actual value provided, not just for access to software. And Agentic AI is the disruption behind this, which could leave them lagging behind more agile, horizontal, AI-first competitors.

The Future is Agentic

As the limitations of SaaS become more apparent—through excessive costs, coordination complexity, and data silos—the rise of Agentic AI marks the next generation of enterprise software. Companies need no longer settle for seat-based, fragmented tools that have ended up creating more complexity than they solve.

Agentic AI is the new frontier: one where automation works end-to-end, across functions, and where businesses only pay for real usage or outcomes delivered. Enterprises that recognize this will see clear rewards—greater efficiency, tangible RoI, and a new era of intelligent, autonomous automation.