The end of SaaS as we know it isn’t hype — it’s a quiet redesign

Agents will sit on top of apps, turning “software use” into automated execution.

Article link: https://open.substack.com/pub/johanosteyn/p/the-end-of-saas-as-we-know-it-isnt?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Audio summary: https://youtu.be/Hm6C924YGj4

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

Most business leaders don’t wake up hoping for yet another software platform. They want work to move faster, with fewer handovers, fewer errors, and less clicking through screens. That is why a provocative idea is gaining traction: AI could “eat” business software by sitting on top of it. Instead of people navigating menus in CRMs, HR systems, finance tools, and service desks, AI agents could take instructions in plain language and execute the workflow across systems.

That is the core premise behind the argument that AI threatens to upend the software stack businesses have relied on for years. The key point, though, is not that SaaS disappears. It is that SaaS becomes less visible, while the underlying platforms become more important than ever.

CONTEXT AND BACKGROUND

Software-as-a-service (SaaS) has dominated business IT for a simple reason: it standardised processes. You didn’t need to build your own CRM or ticketing system; you subscribed to one, trained staff, and enforced a common way of working. Over time, those systems became deeply embedded, not just as tools, but as systems of record: the place where the truth lives, where approvals happen, and where audit trails sit.

AI agents challenge this by inserting a new control layer: If an agent can read your policies, access the right data, and call the right APIs, it can complete tasks without a human opening the application. This is why business conversations are shifting from “should we use agents?” to “how do we deploy them safely at scale?”

The vendors are responding quickly. Microsoft is explicitly positioning agents as something organisations must operationalise with governance, security, and lifecycle management, not just experimentation. Meanwhile, enterprise platforms are retooling their ecosystems around agent innovation and marketplaces.

INSIGHT AND ANALYSIS

For business leaders, the most important shift to understand is that the user interface is no longer the main battleground. In the SaaS era, vendors competed on dashboards, features, and workflow screens. In the agent era, value moves to what sits underneath: data models, permissioning, workflow engines, and auditability. If an agent is going to act on your behalf, the business needs to know three things at all times: what it did, why it did it, and who approved it.

This is why “SaaS becomes infrastructure” is not a throwaway line. The systems of record do not vanish. They become even more critical because they provide the rails that agents run on. When leaders say they want “automation”, what they actually want is dependable execution inside controlled systems. The moment execution becomes automated, the cost of a small mistake can scale across thousands of transactions, tickets, or customer interactions.

There is also a commercial reality emerging: when agents do the work, vendors will be tempted to redefine pricing and value. Seat-based licensing was built for humans. Agents are not humans, but they can generate huge volumes of activity. That creates a new kind of cost and risk management problem for CIOs and CFOs, especially as vendors race to embed agent capabilities into core products. It is telling that AI-driven restructuring is now appearing in the same breath as agent product strategy in the enterprise software world.

IMPLICATIONS

If you are leading a business, the practical question becomes: what must be true for agents to sit on top of our apps safely? Start with identity and permissions. Agents should inherit least-privilege access, with clear approval steps for high-impact actions. Then insist on logs. If an agent touches finance, HR, procurement, or customer decisions, you need traceability that stands up in audits and disputes.

Next, prepare for a hybrid world. Even technology leaders are warning against simplistic narratives that SaaS is “dead”; the more realistic future is that established platforms evolve into end-to-end, agent-governed ecosystems. That means your architecture decisions today should favour clean APIs, clear data ownership, and portability, so you are not trapped by a single orchestration layer.

Finally, treat this as a workforce redesign, not only a tech upgrade. When agents reduce the need for manual navigation, people shift towards exception handling, customer judgment, and oversight. That requires new training and new performance measures, or you end up with automation that looks impressive but quietly erodes accountability.

CLOSING TAKEAWAY

SaaS is not about to collapse. But SaaS, as we have experienced it, humans clicking through screens all day, is being demoted. Agents will increasingly operate the software for us, and that makes the invisible parts of enterprise systems the real differentiators: data integrity, permissions, audit trails, and controls. Business leaders should welcome the productivity potential, but only with a clear-eyed governance mindset. The quiet redesign is already underway. The winners will be the organisations that treat the agent layer as operational infrastructure, not a shiny add-on, and build for safety, accountability, and resilience from day one.

Author Bio: Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net