What kills SaaS isn't AI, but Agents

Author: David Ondrej

Translation: Deep潮 TechFlow

Introduction: Last Monday, Anthropic released a set of plugins for Claude Cowork. It’s not a new model, nor an upgrade to the chatbot, but plugins. Within 24 hours, the software stock market value evaporated by $285 billion. A single-day announcement in the plugin market wiped out more wealth than most industries produce in a year. Wall Street is no longer afraid of AI; they’re afraid of what AI might replace.

David Ondrej points out that what is dying is a very specific type of software business—SaaS built on a weak moat of “beautiful UI + integration + per-seat charging.” When AI agents can perform tasks directly within existing systems, you no longer need 15 SaaS tools with pretty dashboards. Value is being pulled upward into the agent layer and downward into the data layer. Everything in between is being squeezed out.

Full text below:

Last Monday, Anthropic released a set of plugins for Claude Cowork.

It’s not a new model, nor an upgrade to the chatbot, but plugins.

Within 24 hours, the market value of software companies lost $285 billion.

A single-day announcement in the plugin market wiped out more wealth than most industries produce in a year.

Wall Street is no longer afraid of AI.

They’re afraid of what AI might replace.

Two

Here’s where most people misunderstand.

They hear “SaaS is dead” and think it means companies will stop buying software.

That’s not the case, not at all.

What is dying is a very specific type of software business—if you understand which type, you’re looking at the biggest startup opportunity in a decade.

Let me explain:

Three

Over the past 15 years, the SaaS script has been simple:

Find a common business workflow. Build a beautiful UI around it. Add some integrations. Charge per seat, monthly. Defend your position through switching costs and small product tweaks.

This script has created hundreds of billionaires.

But it has a fatal flaw that no one talks about.

Most of the value has never been in the software itself. It’s in the workflow within the software organization.

UI is the middleman.

And AI has just made the middleman obsolete.

Four

Here’s what Anthropic is actually doing—because the headline missed the point.

They didn’t build a better chatbot. They turned Claude into an execution layer.

Cowork plugins allow AI agents to log into your existing tools—your CRM, your documents, your databases—and autonomously execute entire workflows. Legal audits, sales pipeline management, data analysis, production-level coding, and more.

No human involvement needed. That’s the part that’s causing market panic.

Because if AI agents can do work directly within your existing systems—why do you still need 15 different SaaS tools with pretty dashboards?

Here’s the part that should really keep SaaS founders awake at night:

If 10 AI agents can do the work of 100 employees, you no longer need 100 Salesforce seats.

AI doesn’t directly kill software. It kills the number of employees using software. It kills the per-seat revenue model. It kills the business.

Five

This is what I call “Thin Middle Squeeze.”

Imagine three layers:

Top layer—AI agents. The actual executors.

Middle layer—SaaS UI. Dashboards, workflows, buttons you click.

Bottom layer—Record systems. Databases, CRMs, ERPs storing real data.

Now, value is being pulled upward into the agent layer and downward into the data layer.

Everything in the thin middle layer is being crushed.

That’s why Adobe’s forward P/E ratio dropped from 30 to 12. ServiceNow from 67 to 28. Not because people don’t need what they do—because investors realize that when AI agents can bypass UI entirely, the moat around “beautiful UI + integration” is as thin as paper.

Interfaces used to be the product, but now they’re just a shell.

Six

But this is where the “SaaS is dead” crowd is completely wrong.

SaaS isn’t dead; the easy SaaS moat is dead.

That’s a huge difference.

Companies will spend more on software this year than ever before. Enterprise AI capital expenditure alone will surpass $470 billion by 2026. This isn’t a shrinking market—it’s a market exploding in scale.

The money hasn’t disappeared; it’s just moving.

Most people are panicking and completely missing where it’s going.

Seven

Here’s where the money is actually flowing:

• AI Platform Subscriptions

Based on usage. Based on consumption. Not per seat. Companies will pay for AI capacity just like they pay for cloud compute—based on what they use, not how many people sit in the building. This is already happening. GitHub’s AI agents are gated behind a tier with usage-based pricing. That’s the template.

• Record Systems

Agents don’t eliminate the backend; they operate the backend. CRMs, ERPs, data warehouses—these become more valuable, not less. Because AI agents need clean, authoritative, trustworthy data to act on. Garbage in, garbage out. Companies with large, standardized data will win.

• Security, Governance, and Compliance

When agents act at scale, errors happen at scale. Every company deploying AI agents will pay for permissions, audit logs, policy enforcement, monitoring, and evaluation. Boring infrastructure—this will quietly print money over the next decade.

• Results-Based Pricing

No longer “$99 per seat per month,” you’ll see “$5 per contract review,” “$2 per support ticket resolved,” “$10 per qualified lead.” Software priced like labor—because it’s replacing labor. This is where the entire industry’s pricing model is shifting.

• Services

This is surprising. But as building software becomes cheaper and easier, companies will try more custom software services. Implementation, workflow design, migration, integration—demand for services is about to explode. Vibe coding makes creation easy. Making it work in real business is a completely different story.

Eight

So if you’re building a startup—or thinking about it—here’s the only question that matters:

Where are you in the stack?

If you’re building in the thin middle layer—building pretty UI on top of others’ data, charging per seat, with no proprietary advantage—you have serious problems. Not because your product is bad, but because the economics of your position are collapsing in real time.

Your agent layer above is swallowing your interface.

Your record system below is swallowing your lock-in.

You’re being squeezed from both sides—and this squeeze will only accelerate from here.

Nine

Here’s what you should be building:

Build in the agent layer. Create AI-native tools that do more than just display information—they execute workflows. Don’t show dashboards to users. Do the work for them. Price based on results, not seats. Be the thing that takes action.

Build in the data layer. Own proprietary data. Build record systems for domains that don’t yet have good systems. Make yourself the authoritative backend every AI agent needs to insert into. Agents come and go—data layer is forever.

Build infrastructure. Security. Monitoring. Evaluation. Governance. Compliance. Tools to deploy AI agents safely at scale. Not sexy. Extremely profitable. Demand has not even started.

Build services. Help companies implement, customize, and operate AI systems in their actual business. That’s where most of the real-world complexity lies—and where huge value will be created over the next five years.

Ten

Here’s the ironic part no one talks about.

Anthropic’s Cowork—allegedly the product killing SaaS—is itself a SaaS product, sold via subscription online to organizations.

SaaS as a delivery mode has always been good.

But SaaS as a business strategy built on shallow moats and commodity workflows priced per seat—that’s over.

Conclusion

Everyone is watching the $285 billion loss, seeing disruption.

But I see a shift.

That value hasn’t disappeared. It’s moving—from companies capturing value by being the middleman between humans and their tools, to companies capturing value through execution, data, and infrastructure.

The old script was: build workflow UI, charge per seat, grow revenue by increasing employee count.

The new script is: build things with data, execution results, or protected systems. Price based on delivered value, not on seats.

If you’re a founder reading this, the worst thing you can do is panic.

The second worst is to keep building as in 2019.

The best thing you can do is understand where value is moving—and stand where it lands.

The SaaS era isn’t over.

The era of easy SaaS is over.

Honestly, for anyone truly building real things, this is the best news in a decade.