AI agents are redefining SaaS: Monolithic apps lose, operational context wins

The technology market is currently gripped by a provocative narrative: the "death of SaaS." As AI agents evolve from experimental novelties into autonomous workers capable of executing multi-step tasks, the traditional subscription software model is under significant pressure. Investors are questioning whether AI will render established workflows, per-seat licensing, and rigid user interfaces obsolete.

However, this perspective reveals a critical distinction. AI is finally liberating the core business processes and operational data that form the foundation of the modern enterprise, accelerating the collapse of the siloed applications that have hindered them for decades. The "death of SaaS" is not a funeral for software; it is a reckoning for the narrow-minded and boxed implementation of these processes in siloed systems. If AI can generate any piece of code, dashboard, or workflow you need, there is no longer a reason to build new processes inside a single, rigid ERP. In this new era, governance and management must live at the process layer, not the system layer. And operational context, based on process intelligence, becomes the critical enabler that makes enterprise AI actually work.

The collapse of the monolithic stack

For years, the enterprise was defined by "better" versions of specific tools—replacing an old CRM with a cloud-based one or migrating a legacy ERP to a SaaS model. These systems were designed for a world where workflows changed slowly and people were the primary operators.

AI is accelerating the shift away from these monolithic stacks toward composable, modular enterprise architectures. In this world, adding a new process or changing a workflow is as simple as defining it, allowing the enterprise to re-compose itself instantly to meet market demands. Legacy platforms are struggling because they were built for rigidity, not fluidity. AI agents, by contrast, thrive on flexibility.

This creates an economic dilemma for incumbent vendors and a massive expansion of opportunity for those that enable Enterprise AI. An agent in the near term is a multiplier; it allows two senior experts supported by agents to do work that is usually done by ten people. Because these agents can do the work of multiple humans more efficiently, enterprises will require significantly fewer seats. This is why legacy vendors are under pressure: their future revenue is shrinking as AI replaces the features and capabilities that once made their applications essential.

Why agents need an intelligence layer

The disruption of SaaS by AI is a profound and undeniable shift that clears the way for a more important story: enterprises simply can’t scale AI without an operational context based on process intelligence.

By operational context, we mean the ground truth of how the business actually operates—what is done under what circumstances, and which decisions and actions lead to good or bad outcomes. Without this, AI agents provide generic outputs at best and incorrect outputs that harm business interests at worst.

We see this today in software development, a frontrunner of agentic adoption. A senior engineer can use an agent to generate code for simple components, but they must provide precise instructions and manage the system design. The same is true for Finance or Supply Chain. To instruct agents at scale, you need the process context to encode the work that needs to be done. Unlike putting an agent on top of a "dumb" data lake and telling it to "figure it out," process intelligence provides the agent with precise instructions based on how similar cases were best resolved in the past. It takes away the ambiguity, allowing the agent to act with certainty.

Regardless of which AI agents a company chooses to deploy, they need process intelligence to scale. Without it, AI remains an isolated experiment; with it, AI becomes industrialized, controlled, and capable of driving continuous value.

From experimentation to execution

We are seeing demand accelerate as enterprises move from AI "proofs of concept" to actual execution. Celonis customers have already realized more than $10 billion in measurable business value. They are using us to orchestrate agentification, automation, and recomposability across their entire enterprise. For example:

• Deutsche Telekom AG: Saved millions in revenue by proactively engaging at-risk customers and preventing bad customer journeys
• Fujitsu: Reduced excess inventory by 20% by providing buying teams with AI-driven recommendations
• Mercedes-Benz Group AG: Improved on-time delivery, accelerated decision cycles, and made efficiency gains across more than 30 global production plants
• Uniper: Realized double-digit millions in savings across 27 processes by orchestrating an AI maintenance agent
• Vinmar: Increased operational productivity by 20% for a $3B business unit by orchestrating AI agents

A vision for the positive future state

The enterprise of the future is AI-driven and composable—one that can improve continuously, adapt instantly, and innovate freely. In this world, software becomes ambient, and intelligence becomes the primary driver of competitive advantage.

AI makes it possible to re-imagine and re-invent processes so quickly that implementing process changes is no longer a bottleneck to digital transformation. And the more enterprises invest in AI, the more they require the deep operational context that only Celonis can provide. AI is the ultimate tailwind for process intelligence. And along with our partners, we are the architects of the shift toward agents, providing the connective tissue that allows those agents to perform at levels once thought impossible.

When processes work, Enterprise AI works. And when Enterprise AI works, the entire organization thrives.