From Seats to Agents: AI’s Quiet Disruption of the Software as a Service (SaaS) Industry

The Two AI Narratives

The only topic that is as polarizing as U.S. politics today is artificial intelligence (AI). In one corner, proponents argue AI will be the greatest technological revolution ever, eclipsing the big four of the past 25 years: the PC, internet, mobile and cloud. Maybe bigger than all four combined. 

In the opposing corner, some fear AI is going to destroy the world, creating a “Skynet” scenario. 

For me, as a Gen Xer, this is part of our cultural upbringing from the Terminator franchise (1984, 1991 and 2003 films). In these movies, Skynet is the defense AI, an autonomous system that becomes self‑aware and determines humans are the threat, triggering nuclear war via “Judgment Day.” Today, in 2026, certain AI skeptics worry that we are creating the platform for Skynet to happen. 

Of course, AI replacing jobs is another major concern and not just science fiction. There is a school of thought that AI will cause 10%+ of the workforce to lose their jobs, particularly in the “knowledge economy,” where many repeatable tasks can be done by AI “agents.” Agents are software systems that can interpret goals, make decisions and take actions autonomously to complete tasks. Think about research for a court case that used to be done by teams of paralegals in huge file rooms, now done with a trained AI legal assistant. 

But I’m an AI optimist. Every major technological revolution—not just the PC, internet, mobile and cloud, but also electricity, railroads and automation—has ultimately led to more jobs, not fewer. Will it mean we don’t do anything? No. Will it mean we do things differently? Absolutely. In practice, I think the “AI is destroying jobs” narrative may be overhyped, with knowledge workers instead becoming far more productive. 

From Theory to Practice

New roles are already emerging to implement, govern and scale AI across organizations, reducing mundane tasks and allowing people to focus on higher‑value work and deeper client relationships.

At Wespath, we are rapidly adopting AI agents across the organization, starting with simple language models for research, using AI agents as digital personal assistants (mine is named Buford, from the 70’s TV series Buford and the Galloping Ghost, my colleague Karen’s is named Paulie after her French Bulldog) and working to add agents that automate complex, multistep workflows across the enterprise while maintaining the security of our IP and client data. 

Why SaaS Pricing Models Are Under Pressure 

One thing is for certain: AI will change business models and create big winners and losers. 

From my seat, there are already some clear examples of how AI is going to disrupt specific industries, with software as a service (SaaS) at the top of the list in 2026. This is where the conversation moves from productivity and jobs into economics and incentives. Think about some of the enterprise‑wide software services your organization uses, which usually charge on a per‑seat basis. You have 20 people in a division with 20 seat licenses, and the SaaS firm makes more revenue per seat without adding much variable cost as seats are added. All 20 people need the software as part of the team, creating a sticky revenue stream, especially as software is frequently difficult to transition to another provider.

In my days working at banks, managers accountable for profitability within their divisions were constantly asked to add “operating leverage,” defined as how much a company’s profits move when revenues change, driven by how much of its cost base is fixed.

If a business has high fixed costs and low variable costs, it has scale and can create strong operating leverage. Each new client adds more revenue but very little incremental expense, creating more profits. What a great business model! Think about it like an elevator: once you’ve paid to run the elevator (fixed costs), every extra passenger going up adds a lot of upside. But if passengers stop coming, you still pay to run the elevator. That’s the underlying risk. 

Software is one of the best examples of this today. It is a scalable, high operating leverage business, and it can be very profitable, with companies like Microsoft building moats over time and creating high fixed costs for new entrants. 

So, how does this relate to AI in 2026? 

Let’s look at how AI is forcing software companies to rethink their pricing models. First, an example of how the traditional per‑seat model is under pressure right now. Instead of your company paying for 20 software seat licenses, imagine paying for one, with your AI agent sitting on top of it and delivering productive value to the other 19 people on the team. Under a pure seat license model, 95% of the revenue disappears. 

Software companies will push clients toward using their native AI models. Salesforce, for example, has developed its own AI tools, including “Einstein” and “Agentforce.” But are these native tools always a great option? In my view, If I have an agent for myself and for my team, I want that agent to have knowledge of everything, not just Salesforce. 

To adapt, companies like HR software provider Workday are creating models where each client AI agent is treated like a member of the workforce, effectively becoming its own “seat.” These agents are role‑based, audited and governed with strong oversight. What gets lost in human seats can, in theory, be made up for in agent seats. The Workday versus Salesforce examples were discussed on last week’s All‑In podcast, which was the inspiration to write a more digestible blog for the less tech‑savvy. 

Salesforce CEO Mark Benioff, who has a track record of adapting to structural change, is pushing toward what he calls a “headless” model, where systems are accessed directly through APIs (application programming interfaces). Client AI agents query data, upload records and trigger workflows, turning Salesforce into infrastructure rather than just a system of record. In this agent‑first model, Salesforce can charge for activity and usage, not just seats. 

Where Pricing Goes Next

Software pricing is likely to move away from pure seat‑based models toward one or more of the following: 

• Charging to connect to APIs
• Charging for agents as if they were seats
• Charging for usage (searches, data pulls, analytics, etc.) 

And of course, software is just one example from a long list of industries that will be reshaped. Without sounding sensational, I find myself increasingly in the “this may be bigger than the other four tech revolutions combined” camp as Wespath and my team adopt these tools in our daily work (to the point where I honestly can’t fathom taking these agent tools off our desks!). This feels different from prior shifts—you could slow‑walk moving to cloud services because internal servers still worked. This is moving faster, and it is already changing how work gets done. 

I don’t pretend to know exactly where this ends, or which vendors will be standing five years from now. But from where I sit, this isn’t theoretical anymore. The tools are here, they work, and they are already changing business models and workflows in real time. 

For companies, the real risk is no longer getting AI wrong; it’s waiting too long to engage with it. The right posture now isn’t fear or hype, but informed optimism: learn the tools, develop appropriate controls and procedures, experiment early and let them compound alongside your people. 

This is not an offer to purchase securities or an investment recommendation. Please see the Investment Funds Description – I Series and Investment Funds Description – P Series for information about the Wespath funds.