Summary: In this article, AI Strategist Derrick Cash breaks down the rise of Service as Software (SaS), why the familiar seat-based SaaS model is under structural pressure, and what the leaders who get ahead of this shift are doing differently.

Why the SaaS Industry Is Rewriting Itself – and What Leaders Must Do About It

Over the last decade, organizations did exactly what they were told to do.

They adopted best-in-class SaaS platforms, modernized their technology stacks, attended the conferences, retained the consultants, and signed the enterprise agreements. The stack is modern. The processes are documented. The dashboards are live.

And yet, across boardrooms and leadership teams, a different conversation is quietly emerging:

Why aren’t we seeing the outcomes we expected? Why does work still feel slow? Why are costs still rising?

The model that defined enterprise technology for two decades – sell software, let humans operate it, measure adoption – is being disrupted. Industry analysts, including Forrester, have begun describing this moment as a “SaaS-pocalypse.” How software creates value is being rewritten. If you lead teams, manage technology investments, or are accountable for organizational outcomes, this is not merely a trend to monitor. It is a strategic shift to lead.

SaS Vocabulary: #GoalEngineering

This concept will change the way you lead in the age of AI.

AI systems optimize toward defined goals and execute based on available data with remarkable speed and consistency. The challenge is that defining those goals well – with enough precision, balance, and foresight – turns out to be the hardest work in the room. Goal engineering is the deliberate, structured process of defining what AI systems are accountable for producing. This is executive-level work, and it is the SaaS-to-SaS leadership requirement.

When goals are poorly designed, misaligned, or incomplete, an AI system scales the wrong outcomes – faster and more efficiently than any human team could. The technology amplifies the quality of the direction it receives.

Effective goal engineering requires leaders to:

• Define measurable outcomes – not just activities or outputs, but the specific results that matter to the organization, stated with enough precision that AI systems can optimize toward them, and humans can evaluate whether they’re being achieved
• Balance competing objectives explicitly – speed versus accuracy, cost versus quality, automation versus human oversight – and document those tradeoffs so they can be revisited as conditions change
• Establish guardrails – the boundaries within which AI systems are authorized to act, and the thresholds that trigger escalation to human judgment
• Monitor system behavior as an ongoing leadership responsibility – reviewing outcomes, identifying drift, and iterating on goal definitions as production reveals real-world complexity

Execution is no longer the hardest problem. Defining the right outcome is. That definitional work – done with rigor and revisited regularly – is what separates organizations that realize the promise of AI from those that are perpetually disappointed by it.

SaS: Outcomes over Interfaces

The value that lived in the SaaS interface migrates to the intelligence layer.

This one sentence carries most of what you need to understand about where enterprise technology is heading. The user interface — the defining artifact of the SaaS era, the thing vendors invested billions to make intuitive and sticky — is becoming less central to the value equation.

When an AI agent can navigate systems, execute workflows, and deliver results, the human-facing interface becomes increasingly less important. What remains is the outcome. And that changes everything about how technology is sold, priced, and governed.

Traditional SaaS pricing is built on seats — how many people access the software. In a world where AI handles execution, that model is giving way. Microsoft, Salesforce, and ServiceNow are each moving toward consumption-based and outcome-based pricing for their AI product lines. The vendor now delivers the result and, in doing so, shares execution risk with the customer. Vendors who once provided access are being repositioned as operators – accountable for performance, not just availability.

The relationship shifts from “we give you software” to “we produce the outcomes you care about.”

That accountability shift in the vendor relationship is a leading indicator of a much larger shift inside organizations. If the vendor is now accountable for outcomes, someone inside the organization must be equally accountable for defining them with enough clarity that the vendor can be held to them. That responsibility doesn’t exist yet in most organizations. Creating it is the goal engineering imperative.

SaaS Solved Access, Infrastructure, and Cost. Execution Remained Human.

For readers who want the full context on how we got here, this section covers the SaaS era, what it accomplished, and where its limits became visible. If you’re already fluent in this history, the next section picks up where it matters most.

WHAT SAAS GOT RIGHT, AND WHERE IT FELL SHORT

Before the cloud era, deploying enterprise software was capital-intensive and operationally fragile. Organizations purchased hardware, built data centers, and absorbed the full cost of keeping systems running. A new ERP implementation could take 18 to 36 months. Scaling meant buying more servers. Microsoft’s transition to Azure and Microsoft 365, and Salesforce’s reinvention of CRM as a cloud service, demonstrated that mission-critical software could be delivered on demand, updated continuously, and accessed from anywhere. For many organizations, the shift to SaaS reduced infrastructure burden by orders of magnitude. IT departments that spent most of their capacity keeping the lights on could redirect their capacity toward strategy, improving security posture, and increasing user value.

But SaaS left one critical thing unchanged: humans still had to do the work. SaaS improved access and reduced infrastructure overhead. The work itself — the decisions, the processes, the judgment calls – remained human.

Over time, SaaS adoption compounded into a new problem. Today, enterprises commonly operate 20+ tools simultaneously. Employees became the integration layer — manually moving information, reconciling data across systems, context-switching all day. Industry benchmarks suggest 30–40% of SaaS licenses go largely unused in any given month. And the cost of maintaining integrations, managing vendors, and keeping employees trained across an ever-expanding stack is largely invisible – until it isn’t.

The SaaS model that promised efficiency has, in many cases, created a more complex operational burden than the one it replaced. And the application layer most vulnerable to what comes next is not the systems of record — it’s the interface-dependent tools built around human interaction.

Adobe illustrates this tension with precision. Creative Cloud has historically competed on interface mastery – tools designers learn to operate over the years. Adobe’s response has been to embed Firefly AI across Photoshop, Illustrator, Premiere, and Acrobat, shifting the value proposition from “access to professional tools” toward “AI-accelerated creative outcomes.” As VentureBeat noted, Adobe is now asking its customers to let an AI agent handle more of the translation from creative vision to finished output – trusting that the human role shifts from operating tools to directing outcomes. Companies that make this move proactively are repositioning ahead of the disruption. Those who don’t are exposed.

The SaaS Breaking Point: Tool Saturation Without Outcomes

“We bought the software. Why didn’t we get the result?”

This question is now playing out in budget reviews, board conversations, and strategy retreats across every industry. It deserves a direct answer: the result didn’t come because the software was never designed to produce the result. It was designed to give humans better tools to produce the result themselves. What’s changed is that buyers are no longer willing to accept that arrangement – and the technology now exists to offer something different.  A quantum economic shift in action here – now that we see the possibility, it’s becoming real.

Enter Service as Software

A new model is emerging, increasingly referred to as Service as Software (SaS). In the SaaS model, you purchase access to software, your team operates it, and your organization produces the result through human effort applied to better tools. In the SaS model, you define the outcome you need, AI systems execute the workflow, and the result is delivered – with humans managing exceptions. The accountability structure shifts. Execution moves from the customer to the provider.

What this looks like in practice: Customer Support

Under a traditional SaaS model, platforms like Zendesk or Intercom provide infrastructure. Human agents read tickets, classify issues, search knowledge bases, and draft responses. Performance is bounded by staffing and human throughput. Scale means hiring. Under a SaS model, Salesforce’s Agentforce, Microsoft’s Copilot Studio, and Google’s Contact Center AI ingest the request, classify intent, retrieve data, deliver a response, and escalate only the cases that require human judgment. The operational result: faster resolution, lower cost per interaction, continuous availability. The strategic shift: from “help our team work faster” to “resolve the issue.”

That same reorientation is playing out across customer onboarding, procurement, compliance monitoring, document review, and financial reconciliation. The pattern is consistent: AI handles execution, humans govern the exceptions, and value is measured in outcomes rather than seats occupied.

Agentic AI Created the Threshold for SaS

This transformation is being accelerated by the convergence of three forces – and the ripple effect of AI across all three is what makes this moment structurally different from every prior cycle of enterprise technology change.

Microsoft’s Copilot agents, Google’s Gemini platform, Salesforce’s Agentforce, and OpenAI’s operator-mode capabilities have moved agentic AI from research prototype to enterprise deployment at scale. These are production systems operating in regulated industries, handling customer interactions, processing documents, and making recommendations that inform consequential decisions. The capability gap that once made automation impractical for complex work has closed faster than most predicted.

The economics have also moved decisively. The cost of AI inference has been dropping roughly 10x per year, making automation viable at scale in workflows where it wasn’t rational 18 months ago. Rising labor costs and persistent talent constraints are compressing the analysis further. Organizations that once justified high-touch human processes because automation was cost-prohibitive are running the numbers again and arriving at different conclusions.

And buyer expectations have structurally shifted. The enterprise buyer who once evaluated software on features and roadmap now demands measurable outcomes and direct ROI. Deloitte’s 2026 SaaS + AI Agents report predicts that by 2027, the majority of new SaaS contracts will include usage-based or outcome-based components. Bain & Company’s analysis warns that vendors who fail to transition pricing models within 18 months of first seat compression will face permanent revenue erosion.

Yet the gap between capability and deployment remains wide. Agentic AI deployment has crossed 50% among enterprises, but as much as 90% of investment remains focused on tools rather than on the governance structures required to operationalize them. Tools are scaling faster than trust. That gap is the defining tension of this moment — and it is where the ripple effect of AI is most visible, for now.

Who Thrives and Who Dives When AI Lowers the Intelligence Layer

As this transition accelerates, the pattern of winners and losers is becoming clear. The determining factor is not size or incumbency – it is how much of an organization’s value lives in the interface versus the outcome.

The through-line: interface dependency creates vulnerability, and outcome accountability creates resilience. Klarna’s decision to replace Salesforce’s CRM with its own AI system was an early, public signal of this dynamic. When AI makes it rational for a company to build rather than subscribe, seat-based vendors face permanent pressure. Global SaaS spending is still projected to grow from $318B in 2025 to $576B by 2029 – the market is restructuring, not disappearing. But the value is migrating, and it is migrating toward the intelligence layer.

How to Transition from a SaaS Operating Model to SaS Innovation

Service as Software describes the destination. Getting there is a continuous operating discipline – not a project with a go-live date. The organizations navigating this well are treating transformation as an ongoing capability, not a one-time implementation.

In practice, the transition work looks like this:

• Conduct honest automation readiness assessments before committing to deployment – evaluating processes not just for technical feasibility but for organizational and cultural readiness
• Prioritize high-impact, manageable-risk workflows as the proving grounds for agentic AI – where the cost of failure is recoverable, and success builds organizational confidence to move to more complex automation
• Integrate AI agents across existing platform investments rather than replacing systems wholesale – Microsoft Copilot into 365 workflows, Salesforce Agentforce into CRM processes, Google Duet AI into Workspace. Augmentation first, transformation as trust is built
• Establish outcome definitions and governance structures before deployment begins – so that shared clarity about what success looks like exists before the system goes live, not after
• Create feedback loops that connect system performance to leadership decision-making – so what’s working, drifting, or needs to change is visible to the people with authority to act on it

Transformation is earned through iteration, not declared through implementation. Each deployment is a learning exercise. Each learning exercise is an input to the next decision.

AI-Driven SaS: Surface Outcomes, Which Endure

The interface is receding. The intelligence is moving to the layer below what users see. What surfaces – what the organization and its customers actually experience – is the outcome.

The organizations that will define the next decade are the ones that understand this at the leadership level and act on it with urgency. That means investing in the data foundations AI depends on. It means building the goal engineering capability to define outcomes with precision. It means developing the cultural conditions that allow transformation to land. And it means treating all of that as continuous work rather than a project to celebrate and close.

The new question is: What outcomes are we designing our systems to deliver, and who is accountable for them?

Every leader reading this should be able to answer that question at the levels of their organization, function, and team. If the outcomes aren’t defined, if accountability isn’t assigned, if governance doesn’t exist, the technology investment will underperform regardless of the vendor, budget, or implementation quality.

The interface is disappearing. Clearly defined outcomes and equally polished governance are paramount. The leaders who internalize that shift early will define what comes next. The front windshield of your vehicle is significantly larger than your rearview mirror, by design.  What sorts of horizons will we see when traveling at the speed of AI?  No one really knows.  But we’re still driving, and I’m looking forward to seeing the future seemingly generate before our very eyes.