Enterprise Software (SaaS) • 2 hours ago • Neha Jamwal
For more than two decades, the SaaS business model has been built around a relatively simple equation: more users mean more software seats, and more seats mean more recurring revenue. That model helped transform enterprise software from a capital-intensive product into a predictable subscription business. But the rapid adoption of AI agents is beginning to challenge the assumption at its core. If an AI agent can perform work that previously required several human users to navigate multiple applications, the relationship between software usage, seats and revenue starts to break down.
This is no longer merely a theoretical concern. Gartner estimates that up to $234 billion of enterprise application software spending could be exposed to agentic arbitrage by 2030, representing roughly 20% of enterprise SaaS spending. The research argues that agents can execute tasks across multiple systems while bypassing traditional application interfaces, potentially weakening the link between user growth and software revenue.
The implication is not that SaaS is disappearing. Rather, its economic model is being redesigned.
The Seat Is No Longer the Whole Product
Traditional SaaS pricing made sense when the human user was the centre of the software experience. A sales representative needed a CRM seat, an HR professional needed an HR platform seat and an analyst needed a data platform seat. The interface was where work happened, while the software’s features and workflows justified the subscription.
Agentic AI changes that relationship. An employee may increasingly tell an agent what needs to happen, while the agent interacts with CRM, ERP, HR, procurement and other enterprise systems through APIs. The human may never open several of those applications directly.
IDC describes this shift as a move toward SaaS becoming “featureware” behind an AI agent layer. Its research argues that when agents increasingly handle user interaction and transactions, traditional justifications for per-seat pricing — including the interface, feature breadth and direct user experience — become less important.
This creates a fundamental challenge for vendors. If one employee equipped with an agent can effectively perform work previously distributed across several software users, adding more seats may no longer be the most meaningful indicator of value.
From Software Access to Business Outcomes
The more important shift is therefore not simply from seats to usage. It is from access to outcomes.
Deloitte expects SaaS pricing to increasingly experiment with combinations of subscription, usage and outcome-based models as agentic AI becomes more prevalent. An agent could be priced according to actions performed, computing or API consumption, tasks completed, or eventually the business results generated by those tasks.
That distinction matters because usage alone does not necessarily equal value. An agent that makes 10,000 API calls is not automatically more valuable than one that makes 1,000. A procurement agent that processes thousands of transactions may be valuable because it reduces cycle time and errors, while a customer-service agent may be valuable because it resolves more cases without human intervention.
For SaaS vendors, this creates a new pricing question: what exactly is the customer paying for?
The answer could increasingly be a completed workflow rather than access to a software interface. Gartner’s recent research on agentic arbitrage similarly argues that enterprise buyers are looking for measurable outcomes and that vendors need to shift from interface-based value toward outcome-based value.
Enterprise Software Does Not Become Less Important
There is an important nuance in the emerging SaaS disruption story. The declining importance of the interface does not necessarily mean the underlying enterprise application becomes irrelevant.
In many cases, the opposite may be true.
Enterprise applications contain years of business rules, permissions, workflows, customer information, financial records and operational context. They are also deeply integrated into existing technology environments. Replacing them with an AI-generated application is considerably harder than generating a new interface or automating an isolated task.
That helps explain why the current AI disruption of enterprise software has been more gradual than some of the early “SaaSpocalypse” predictions suggested. Recent reporting has highlighted the resilience of major enterprise software vendors, particularly because large organizations remain dependent on deeply integrated systems that are expensive and risky to replace.
The battleground is therefore moving underneath the interface. The question is increasingly whether a SaaS vendor owns the system of record, business logic, trusted data and execution layer that agents need in order to complete work safely.
The New SaaS Product May Be an Agent
This is where the opportunity becomes more interesting for software vendors.
Instead of simply adding an AI assistant to an existing application, vendors can turn their domain expertise into autonomous agents capable of executing complete workflows. An enterprise software provider with deep knowledge of procurement, finance, HR or customer operations has an advantage because its agent can already understand the underlying data, permissions and business rules.
IDC’s research points toward precisely this model. It argues that enterprise buyers increasingly expect SaaS vendors to supply agents while also providing the trusted data and context that custom-built or third-party agents need. In this model, the application evolves beyond its traditional interface and becomes a governed environment that external and internal agents can safely interact with.
That could produce a very different SaaS architecture. The traditional interface would remain available for humans, but it would become only one access point among several. Conversational interfaces, embedded workflow experiences and machine-to-machine APIs could all sit on top of the same underlying application and business logic.
The strongest SaaS vendors may therefore stop thinking of their product primarily as a destination employees log into. Instead, they may increasingly position it as an execution platform that humans and agents can use in different ways.
Pricing Will Become More Complicated Before It Becomes Clearer
The transition will not happen through a simple replacement of “per seat” with “per task.”
For a period, enterprises are likely to operate with hybrid models. A company may continue paying an annual subscription for the core platform while separately paying for AI-agent usage, additional compute, premium models or completed workflows. Some vendors may bundle a certain amount of AI usage into existing subscriptions and charge additional fees once customers exceed defined limits.
Recent pricing changes illustrate how quickly this experimentation is already emerging. SaaS Price Hub’s September 2026 pricing analysis, for example, highlighted changes in AI-credit allocations alongside stable seat prices, illustrating how the effective economics of a software subscription can increasingly depend on included AI consumption rather than the headline seat price alone.
For enterprise buyers, that means the traditional question of “How much does this software cost per user?” will become less useful. Procurement teams will increasingly need to understand the relationship between licensing, agent activity, consumption limits and business outcomes.
For vendors, meanwhile, the challenge will be even greater. Outcome-based pricing can be attractive because it aligns vendor revenue with customer value, but measuring and attributing outcomes becomes considerably harder when multiple applications, agents and human workers contribute to the same business process.
The Winners Will Own Context and Execution
The SaaS companies most exposed to agentic disruption are not necessarily those with the most users. They are those whose value proposition depends heavily on humans navigating a proprietary interface to perform relatively standardised tasks.
By contrast, vendors with strong systems of record, deeply embedded workflows, differentiated data and sophisticated governance capabilities may have a stronger position.
This is why enterprise SaaS incumbency remains important. The very infrastructure that once made traditional SaaS difficult to displace — integrations, data models, permissions, compliance controls and workflow dependencies — can become an advantage in an agentic environment.
The winning product may therefore be the one that gives an agent the safest and most intelligent way to execute a business process. Its value will come less from how many screens employees use and more from how effectively the platform provides context, controls execution and produces measurable results.
SaaS Is Moving Into an Agentic Economy
The SaaS industry’s next transformation is unlikely to be a clean break from subscriptions. Instead, the market is likely to move through a period of hybrid pricing, changing product architectures and experimentation around what constitutes a software user.
Deloitte expects the transition toward agentic SaaS to be gradual, noting that deeply embedded enterprise applications will take time to replace even as AI-first competitors emerge. Recent market developments reinforce that view: enterprise software is facing renewed disruption fears, but major vendors are simultaneously integrating AI agents into their platforms and adapting their commercial models.
For CIOs and technology leaders, the important takeaway is not that every SaaS contract should immediately be replaced with an outcome-based agreement. It is that the assumptions behind today’s software portfolio are changing.
The next generation of enterprise SaaS will be judged less by how many people log into it and more by how much work it can safely get done.
The seat is not necessarily disappearing. Its role as the fundamental unit of SaaS value is.
Key Takeaways
- Agentic AI is challenging the traditional per-seat SaaS model because agents can perform work previously distributed across multiple human users.
- Enterprise SaaS is shifting from access toward outcomes, with usage-, task- and outcome-based pricing likely to coexist with subscriptions.
- The application interface may become less important as agents increasingly interact with enterprise systems through APIs and workflow layers.
- Data, business context, permissions and execution capabilities are becoming strategic SaaS assets, potentially making established enterprise platforms more valuable in an agentic environment.
- SaaS vendors will need to become agent suppliers, not simply software providers with AI features added to existing products.
- Enterprise buyers should prepare for hybrid pricing models in which seat licenses, AI credits, consumption and outcome-based charges coexist.
- The disruption is more likely to reshape SaaS than eliminate it, creating a new economic model around software-enabled work rather than software access.