Your Stack Can’t Think

I joined Salesforce in April 2000 as one of its early IT hires – employee #70! For nearly ten years, I helped and watched as we solved the last generation’s IT governance challenge – moving from on-prem chaos to centralized, controllable SaaS applications. We gave CIOs visibility and control they’d never had before. But what we’re building now threatens to undo all of that.

Today, we’re hitting the ceiling of that model.

SaaS was built to streamline functions, including sales, HR, finance, and support. And it did. SaaS gave us control: everything happened inside apps we could see and manage. But it was never built to think. Your stack today takes information in, pushes information out. It executes – but it doesn’t understand. It doesn’t adapt. It doesn’t anticipate.

Now something new is emerging. A model where AI operates across systems, surfacing what matters and acting on it. Not bots. Not assistants. Actual execution engines. AI agents break the control SaaS gave us: they work across systems, make decisions, and/or take actions.

This isn’t a minor evolution. It’s a break from how enterprise IT has worked for the last 20 years.

SaaS Is No Longer the Center of Gravity

In the last year alone, nearly every major software provider has repositioned itself around AI. They’re launching copilots, agents, assistants – anything to meet rising expectations. And for good reason. Only 13% of business leaders say SaaS features are central to their AI strategy. Most are looking beyond legacy apps, toward API-first models, orchestrated workflows, and embedded intelligence that works across systems – not just inside them.

The message is clear: Your stack can’t think. But worse: you can’t govern what you can’t see thinking.

From Tools to Outcomes

SaaS exploded because it made it easier to acquire and manage software tools, with much lower upfront costs and faster than on-prem ERP. But tools alone aren’t enough anymore. Businesses want outcomes that scale. Today, the average company uses 106 SaaS applications but that sheer volume hasn’t translated to better outcomes.

Modern enterprises don’t operate in silos. Their work spans teams, systems, and functions. But SaaS locked intelligence inside applications. It created fragmentation that we’ve spent the last decade trying to stitch back together – with middleware, custom logic, and manual workarounds.

Even the AI embedded in today’s apps is stuck in those silos. It can analyze, summarize, even recommend – but only inside its lane. That’s not intelligence. That’s bounded automation. That’s AI stuck at the edge of the enterprise.

Agents Change the Game

Intelligent agents represent a fundamentally different model.

They don’t live inside apps – they live across them. They act based on real-time signals. They work autonomously but stay within policy. And they’re not just here to help – they’re here to get things done.

Imagine procurement agents that preempt disruptions. Support agents that resolve issues before customers complain. Finance agents that course-correct risk before it spreads.

Early results are promising: 66% of organizations adopting AI agents report measurable value through increased productivity.

Legacy SaaS is being redefined. In this new model, SaaS becomes infrastructure. A data layer. The new center of gravity is the agentic layer where decisions get made, actions are taken, and value is realized.

It’s more than a technical shift. It’s a different way of thinking.

We Need Control – Not Just Connectivity

As a CIO, I don’t just care what AI can do – I care what I can govern.

Agents can’t run wild. They need policies, visibility, and approvals. They need to be traceable. Trust is earned, not assumed.

And they need a human-in-the-loop interface – somewhere people can intervene, prioritize, set guardrails, and understand what’s being done and why. The future of enterprise IT isn’t just autonomous. It’s creating agent accountability. IT will become the HR department for agents.   They will need to be onboarded, mentored and managed like human employees.

Most CIOs I talk to have AI projects underway, but few that scale across departments. The pattern is familiar: One-off assistants, trapped in tools, delivering isolated wins. That’s not transformation. That’s a tactical lift. And it won’t last.

Enterprise-scale AI implementations have nearly doubled – from 8% in 2023 to 15% in 2025 – showing early momentum but still limited widespread adoption.

This Is Bigger Than SaaS vs. ERP

I’ve lived through big architectural shifts. ERP to SaaS was one of them. But this one’s bigger – and it’s moving faster. While many organizations are still scaling their AI efforts, the broader transformation is already underway. In just a few years, agentic AI will be embedded in over a third of enterprise applications, driving a meaningful share of everyday decisions without human input. This isn’t a distant vision – it’s unfolding now, and the shift is real and accelerating.

You need to be actionable. And fast. CIOs must consider how to approach this architectural change:

1. Design for agent oversight from day one – don’t just retrofit and move on. Consider:

• • Agent transparency and auditability
  • Cross-system policy enforcement

2. If employees are meant to work alongside agents, treat agents like employees. Onboard them, monitor them, create goals, and set boundaries.
3. Choose platforms that let you orchestrate, not just integrate.

This time we’re not just replacing software. We’re replacing assumptions. Work no longer happens in the stack. It happens through agents.

CIOs Will Decide How This Plays Out

The hype cycle is fading. Now comes the hard part: designing systems that are trusted, accountable, and built for intelligent execution. That means moving beyond disconnected tools – toward one unified intelligent environment where agents and humans collaborate in real-time.

For CIOs, this shift means moving from systems integrator to intelligence architect. It’s no longer about stitching platforms together – it’s about designing a stack that can sense, respond, and evolve. That’s not a tech project. It’s a leadership one.

You can’t retrofit your way into that future. You have to rethink the architecture of work itself.

SaaS won’t disappear. But it will be outpaced. The shift is happening whether you lead it or not. The question is: will you govern it?