Is SaaS Dead? No. But AI Just Killed the Dashboard.

I told my wife about Claude's new Cowork feature — a tool that lets AI enter your software and take real actions on your behalf. She thought for a moment and asked: "Wait, so I can stop logging into Gmail?" I answered: "Exactly. But that's only half the story."

The other half isn't about Gmail. It's about us — the software founders and operators. And it's equal parts terrifying and thrilling.

The Dirty Secret of the Software Industry

Most SaaS products that businesses pay for today are essentially a pretty interface sitting on top of data.

That's it. That's the whole trick.

OK, maybe not just a dashboard. Maybe they threw in some scripts that do "simple" processing along the way, maybe some automation here and there. But let's be honest… there's no real depth here. In most cases, it's just content accessibility wrapped in one form or another.

Someone took a workflow that exists in every company, wrapped it in colorful buttons and a nice dashboard, charged X dollars per user per month, and called it SaaS. This model generated hundreds of billions over the past 15 years. But it always relied on one assumption: that you need a human to click the buttons.

So what happens when AI knows how to click the buttons itself?

The Moment I Realized the Interface Is No Longer the Product

What Anthropic did with Cowork (and what's happening with Altman's Operator and others like them) isn't just another chatbot. They built agents that can enter your CRM, your ERP, and your email and actually do the work. Not "help" you work — actually work.

And the moment an AI agent enters a system, it doesn't need pretty buttons. It doesn't need a polished "user experience." It needs access to data and the ability to execute.

That's the tsunami currently sweeping Wall Street, knocking down the stock prices of massive SaaS companies. Investors have realized that companies whose entire value is "a nice interface on top of someone else's data" are in serious trouble. Their moat, which once looked deep, turned out to be a puddle.

The Middle Trap: Where Does the Money Go?

To understand who survives this, think of the software world as a three-layer sandwich:

1. The Top Layer (Execution): The AI agent. The one that actually does the work.
2. The Middle Layer (Management): Classic SaaS. The dashboards, the buttons, the interfaces.
3. The Bottom Layer (Truth): Systems of Record. The databases, the raw data, the organization's real infrastructure.

What's happening right now is a violent compression of the middle.

Value is fleeing upward (to whoever does the work) and downward (to whoever holds the unique data that makes the work possible).

And what happens to whoever's left in the middle with a pretty dashboard but no unique data and no autonomous execution capability? They get crushed.

Why We're Not Scared at Tracer (Quite the Opposite)

When I look at our space — fleet management — I realize we're in exactly the right place.

At Tracer, we collect tens of millions of GPS data points every day. We analyze driver behavior in real time, fuel consumption patterns, safety violations. This is the raw data. We own "the bottom layer" — ground truth. No pretty interface can fake it, and no generic AI can make it up.

We understood a while ago that our customers don't actually want "the middle layer." Business owners and fleet managers don't wake up in the morning burning with desire to stare at graphs in our dashboard and download spreadsheets. They want to know their employees are doing what they're supposed to do, that there's no fuel theft, that the fleet is in good shape, and that drivers are safe.

That's why the real revolution we're leading at Tracer is skipping the middle entirely: a direct connection between data (layer 3) and execution (layer 1).

Already today, a massive part of the value our customers get happens without a single log-in. The system detects an anomaly? It doesn't wait for the manager to open a laptop, download a spreadsheet, filter, pivot, and discover only what "he knows to look for" because he "understands the business best." It understands what the user needs and sends a precise alert to WhatsApp at 7:00 AM with an analysis of the event and a recommended action. Reports that build themselves, insights that arrive via push — not pull.

We're building a system that works for the customer, instead of the customer working inside the system.

Sounds trivial? Try explaining that to a sales lead who asks: "Wait, but where do I see the cars on the map?" Or my personal favorite: "What's the difference between you and [competitor X]?" Ugh… the old instinct still searches for the screen, the visual interface. But the real value — the big money — is in the autonomous execution happening under the hood. (And yes, we still have all the visual stuff for those wondering.)

The Evolution: From Cloud Native to AI Native

Let me be precise here — it's not like dashboards will die tomorrow morning.

There are complex systems that require human judgment, file uploads, or process visualization (think maps, medical imaging, or video editing). Those aren't going anywhere.

But even they will transform into something new: AI Native.

In these systems, the user's interaction will be directly with the AI. The interface won't be a "control panel" where you click buttons, but a "workspace" where you give the instruction and the AI runs the interface and executes the actions. We won't need to fill out 50 form fields; we'll upload a file, write a sentence, and the system will understand, execute, and surface only what requires approval.

The Future: Transparent Software

So is SaaS dead? Not at all.

What's dead is lightweight SaaS. The SaaS of "let's take a generic process and wrap it up nicely."

We're entering the era of "Transparent Software."

In this era, the best CEOs won't ask their CTOs "What does the system look like?" or "How user-friendly is it?" The questions will be: "How much time did it save us?" and "How many decisions did it make for us?"

Companies that have real, proprietary, unique data (like the tens of millions of kilometers we analyze) and know how to build AI agents that act on that data — those are the ones that will win.

The best dashboard in the world is a screen that's off, because the work has already been done in the background. That's where we're headed. And whoever insists on selling "buttons" instead of "outcomes" will very quickly discover there's nobody left to click them.

Daniel Priscu, CEO & Founder, Tracer Technologies Ltd.