The SaaSpocalypse Isn't About Software. It's About Work.

Last week, roughly $800 billion in software market value got wiped out. The media called it the "SaaSpocalypse," and everyone with an opinion rushed to pick a side.

NVIDIA founder Jensen Huang called it illogical and argued AI will use existing software tools rather than reinvent them. Jason Lemkin, “The Godfather of Saas” and founder of SaaStr, took what I think is the most honest position: this isn't the death of SaaS, but "the final end of easy SaaS." He shared on Lenny Rachitsky's podcast that SaaStr went from ten full-time go-to-market people to a hybrid workforce of 1.2 humans and twenty AI agents, with roughly the same business performance. Jason Calacanis said his top two agent-builders are each worth 200 of his other employees. And others debated profit pools, legacy layers, and outcome-based pricing. 

But here's the thing that kept nagging at me as I listened to all of it: every single one of them is debating what happens to software. None of them are really asking what happens to the workforce that operates inside it.

The Question Nobody Is Sitting With Long Enough

For all the debate about models, platforms, profit pools, and pricing, there’s a more poignant question hiding in plain sight. If software is no longer just supporting work but actively performing it, then the real issue isn’t what tools we buy. It’s how work itself is designed, managed, and measured when part of the workforce is no longer human.

Lemkin's example is a perfect illustration of what I mean. He replaced ten humans with 1.2 humans and twenty AI agents. Same revenue. He even gave the agents names and desks. But the business performance he described was roughly the same as what the humans delivered.

That's a cost reduction story, and it's a meaningful one. But there are questions underneath it that I think matter more: 

• Who’s coaching those twenty agents? 
• Who decides when one is underperforming? 
• Who understands the work deeply enough to know what "good" looks like for an AI agent doing outbound sales for a company like SaaStr? 

Lemkin mentioned that Amelia, his chief AI officer, spends about 20% of her time managing and orchestrating the agents. That's the piece most people gloss over, and it's the whole game.

Calacanis got closer to naming it when he said his top two agent-builders are worth 200x the other eighteen employees. That's not a technology insight, that's a workforce composition insight. The humans who can design, manage, and coach digital workers are exponentially more valuable than the humans who can't. And there aren't very many of them right now.

This is what we call the talent paradox: you can’t orchestrate a workforce you can’t build, and the hardest part of building a hybrid workforce isn't the technology. It's finding and developing the humans who can operate effectively alongside digital workers. Not just people with technical skills, but people who understand the outcomes their role exists to achieve and can translate that into a clear direction for a digital teammate.

What This Looks Like in Practice

I'll share a concrete example from inside Asymbl that I think helps make workforce composition less abstract.

I have a digital worker, Bradley, whose role is to deliver a visual briefing that shows me what happened across the business, what’s progressing well, and where my attention is needed. To do that he coordinates 20+ AI agents across systems like Claude, Agentforce, Slack, Salesforce, Asymbl Recruiter Suite, Gemini, Google Workspace, ChatGPT, and several other SaaS applications and sources. The work is checked and double checked before it reaches me.

This is work that would take two to three people about a month to compile and analyze. I receive the briefing in about fifteen to twenty minutes.

But here’s the part that matters, and that most people skip past: Bradley works because I know the outcomes I want. I know where the information lives, because they’re the same systems I would otherwise check one by one. And most importantly, I’m willing to do the ongoing coaching and management required.

That last part is the real commitment. The technology is table stakes. The orchestration of how humans and digital workers collaborate, that’s where the value is.

This is just one of hundreds of examples of what we’re doing right now at Asymbl to orchestrate work across a hybrid workforce of 150+ digital workers alongside our human team. We’re heads down and busy proving this internally, and we're also helping our customers get started with what I call "validation of belief" projects instead of proofs of concept. These efforts are real, not conceptual. The point isn't to prove the concept works. It’s to deliver enough value that you believe in it, which is what radically changed my perspective a couple of years ago.

The Four Layers That Matter

I recently wrote a piece on the Asymbl blog, building on an analysis by Vernon Keenan that I think frames this well. The way I see it, there are four layers to getting real value from AI in the enterprise:

The first layer is the LLM itself, the raw reasoning capability. The second is technical orchestration, the infrastructure that makes AI work reliably in production. This is what Salesforce is building with Agentforce, what Keenan's analysis focused on, and what Huang is essentially defending when he says AI will use tools rather than reinvent them. The third layer is institutional memory, the understanding of why your organization does things the way it does, not just what data exists but why decisions were made and what customers expect. The fourth is workforce orchestration, the discipline of designing, coordinating, and scaling blended human-digital teams with job descriptions, success criteria, managers, coaching, and measurement.

Most of the SaaSpocalypse conversation is stuck on layers one and two. That's where the billion-dollar companies are focused, and it makes sense for what they do and what they sell. But layers three and four are where the business value compounds, and almost nobody is focused on them yet.

The UI/UX Curveball

There’s one more dimension to this shift that most software companies aren’t prepared for, and it has nothing to do with models or infrastructure. It has to do with the user interface (UI) itself.

Here’s something I've observed as digital workers become embedded in real workflows: sometimes they perform better navigating a software application's interface the way a human would, rather than connecting directly to that application's backend systems. Even when using Model Context Protocol, digital workers often navigate the UI more effectively than they query the API. That interface was built for workflows, and agents can reason through it the same way people do. The interface encodes context, sequencing, and guardrails that direct system connections don't.

That's a curveball for software companies because it raises a question the industry wasn’t prepared for: do UIs matter more in an AI-driven world? If digital workers are increasingly the end users of enterprise software, the design of that software starts to matter in entirely different ways. It’s no longer just about human-centered design. It’s about whether an interface enables or constrains the digital workers now doing the work.

What happens when a digital worker can access any UI through any flow of work? The walls between applications start to dissolve, and the value shifts from the interface itself to the orchestration of how work moves across all of them.

The Real Disruption Is the Workforce, and Software Has to Evolve With It

What the SaaSpocalypse really reflects isn’t the collapse of software, it’s the market struggling to price a shift it doesn’t yet understand. For investors, model performance and technical orchestration layers may determine near-term winners. But for companies trying to generate real returns from AI, that’s not where the leverage is.

The advantage will belong to organizations that can design, coordinate, and scale a hybrid workforce of human and digital workers, and develop the people capable of making that system work. 

That’s the difference between deploying AI and operationalizing it.

At Asymbl, this is the problem we’re focused on solving. We call it workforce orchestration, because the challenge isn’t a feature or a tool. It’s a discipline. One that spans recruiting, digital labor, operating models, and ongoing management of work across humans and machines.

Models will continue to improve, and technical platforms will continue to mature. But what will separate companies over the next few years is whether they’ve built the organizational capability to operate as a hybrid workforce. The talent paradox only sharpens from here. As digital workers scale, the scarcest resource won’t be technology. It will be the humans who know how to design, orchestrate, and coach them.

This isn’t a SaaSpocalypse. It’s a workforce transformation. And the companies that recognize this will capture the value others are still trying to explain.