Artificial Intelligence
3 Reasons Enterprise Software Is the Right Foundation for AI
What makes B2B software feel antiquated is also what makes it the best platform to deploy AI.
Julie Colwell
Principal Strategist
Workday
For the past two years, enterprise software companies have been under scrutiny. If agents can do the work, why keep paying for the software? The initial stock sell-off suggested an imminent “Saaspocalypse.”
But that hasn't happened. Instead, the picture has brightened. Enterprise software companies are proving they are uniquely positioned to unlock responsible AI at scale. And, they’ve already won the hardest battle: earning customer confidence.
There are three key advantages SaaS companies have that position them to be the right foundation for AI:
The operating model (IT): how the system itself is structured—roles, permissions, approval chains, and jurisdictional rules built into the platform.
The pricing model (finance): how work gets measured, attributed, and billed.
The talent model (HCM): how a workforce gets onboarded, scoped, reviewed, and offboarded.
This head start is what makes enterprise SaaS the safer place to let an agent act on someone's pay, access, or a company's books. It's also why the “Saaspocalypse” fear hasn't materialized. The parts of SaaS that felt slow have turned out to be the precondition for trusting AI with anything consequential.
Report
AI Agents Are Here—But Don’t Call Them Boss
1. The Operating Model Is More Than a Dataset
Initially, enterprise software companies tried to hang their advantage on data—more transactions, cleaner records, one system instead of a patchwork. That's true, but it's only part of the story.
Data can tell a model what happened, but it doesn't say what's allowed to happen next, under whose authority, in which jurisdiction, or who's accountable if it goes wrong. These are non-negotiables. Customers can't risk the potential to instantly scale mistakes, train agents wrong, or deploy them without understanding the scope of consequences. The stakes are too high.
Workday CEO Aneel Bhusri has said HR and finance are deterministic by necessity, demanding complete accuracy and auditability across jurisdictions. HR analyst Josh Bersin agrees: the permissions, security model, and compliance logic already embedded in a system of record should act as rails an agent can run on safely. Rebuilding those rails from scratch is expensive and risky in ways that are easy to underestimate from the outside.
In other words, enterprise AI requires a reliable operating model, not just a dataset. It needs to be built, case by case, inside the system of record that runs a business.
It is built, case by case, inside the system of record that runs a business.
Take Medtronic, for example. The healthcare technology company set out to automate 80% of its HR processes, starting with a talent-acquisition function handling roughly 12,000 hires a year. Their AI agent is configured differently by role—a senior medical specialist gets a more nuanced screening path than a transactional hire. This is exactly the kind of role-based, policy-aware scoping that has to be built into the platform, not supplied by the model.
Instead of teaching a general-purpose agent a company's rules by trial and error—and hoping the errors are cheap—the platform already knows what a termination looks like in France versus Texas. The permissions are inherited, not learned.
This doesn’t exist in an external AI model.
2. The Pricing Model Charges For the Work, Not Just Seats
The second advantage enterprise software companies have is their evolving pricing model.
Consumption-based or usage-based frameworks charge for the work agents complete rather than a fixed number of employee seats.
Gartner projects at least 40% of enterprise SaaS spend will move to usage, agent, or outcome-based pricing by 2030. Much of the industry is now using this model or a hybrid version of it.
Usage-based pricing is the only fair way to value a service that is both software and labor.
Usage-based pricing is the only fair way to value a service that is both software and labor. What agents do and the tasks they complete can be measured, attributed, and audited with the same permissions and record-keeping the operating model depends on.
In this model, the invoice is a defensible financial record because every unit of agent work is logged and attributable.
3. The Talent Model Manages Agents Like a Human Workforce
The last two years have shown us that the future workforce includes both employees and agents, and companies that don't learn to manage that mix will fall behind.
Enterprise SaaS is primed to help companies make this transition. B2B software is already embedded in core HR workflows: onboarding, role definition, performance tracking, access changes, offboarding. When AI is strategically integrated, rather than bolted on, the talent model can manage agents alongside human employees, without reinventing the process.
A company can onboard an agent securely, define its role and responsibilities, and track its cost and impact the way it would for a person.
In other words, a company can onboard an agent securely, define its role and responsibilities, and track its cost and impact the way it would for a person. Customers already trust enterprise software companies with these capabilities, making AI from enterprise SaaS companies far less risky than ungoverned solutions that are bolted on.
Why This Is a Welcome Relief
The things that made enterprise SaaS seem traditional—governance overhead, billing discipline, HR bureaucracy—turn out to be exactly what AI needs in order to act safely. A platform that already knows who can approve what, can meter and audit a unit of work, and manage a workforce isn't a legacy system standing in the way. It's the platform built to let an agent act on an organization's behalf.
It's the platform built to let an agent act on an organization's behalf.
We should expect the line between employee and agent to keep blurring inside these platforms, with more roles, more approval chains, more of a hybrid org chart. We should expect pricing to keep shifting toward what an agent actually accomplishes, with the record to prove it. And, we should expect the scope of what agents are trusted to do to widen gradually, function by function, as guardrails are tested.. The vendors who spent years building software for people are the ones best positioned to extend it to agents because a foundation of trust was already there.
Report
AI Agents Are Here—But Don’t Call Them Boss
A gap between effort and impact is the real ROI problem. Discover the solution.