Last March, at a demo day hosted by a San Francisco startup accelerator, one team put up a slide. The title was short: "We don't use SaaS." The audience laughed—and before the laughter had faded, a dozen investors raised their hands. That team walked away with more business cards than anyone else in the room. That is how 'SaaSpocalypse'—the software-apocalypse thesis—became Silicon Valley's most contested word of 2026, and the question has since crossed the Pacific to land on the monthly invoices of Korean founders.

The Claim That Software Has Started Eating Itself

The logic of the SaaSpocalypse is simple. As AI agents begin automating complex workflows, a single agent can now take over tasks that used to be split across separate subscription tools—CRM, project management, customer support, design drafts. If a company can use one AI agent to organize its sales data and send follow-up emails automatically instead of maintaining an annual Salesforce contract, the logic justifying the existing subscription fees starts to wobble. The entire premise of the subscription software market was "we make tedious human tasks easy"—and now agents have started handling those tasks more directly.

Several observations converged to intensify the debate. Agent features from OpenAI, Anthropic, and Google began handling complex multi-step work in the second half of 2025, and the latest Y Combinator batch saw a visible rise in startups positioned around "replacing a specific SaaS category with agents." The share prices of flagship SaaS companies like Salesforce, ServiceNow, and Zendesk kept wobbling every time an AI agent announcement dropped—a sign that investors have started taking this structural shift seriously.

When The Economist covered the trend head-on, the context was that the debate had spread beyond Silicon Valley's insider community into mainstream corporate strategy. Predictions that "30% of the SaaS market will be replaced by agents by 2030" began circulating as conservative estimates rather than radical scenarios—and applied to a global SaaS market estimated at $4 trillion, that is not a number anyone can shrug off. This isn't a passing tech fad; the revenue structure of the software industry itself is being shaken.

The shift feels close to home in Korea as well. A large share of Korean startups and solo entrepreneurs run on some combination of Notion, Slack, Figma, HubSpot, and Zapier. SaaS subscription stacks running 150,000 to 500,000 won a month (roughly $110 to $370) are not unusual. Add team collaboration tools, accounting software, and email marketing platforms, and some businesses spend millions of won a year. The claim that this entire stack could be replaced by an integrated agent within six months to a year lands as real pressure to re-examine fixed costs.

But SaaS Won't Disappear as Fast as You Think

There is also considerable skepticism about this outlook. The claim that AI agents will replace SaaS in the near term rests on several hidden assumptions, critics point out—and not all of them hold.

The most frequently raised counterargument is structural switching costs. SaaS tools don't just provide features. Notion holds years of accumulated document structures and team collaboration habits; Slack holds channel histories with customers and partners; HubSpot has accumulated deal records and sales pipeline data. Migrating all of this to an AI-agent-based system is not simply swapping software. It cascades into data migration, team training, redesigning work practices, and adjusting how you collaborate with clients. That cost often far exceeds many months of subscription fees—which is why the simple math of "cancel and save" doesn't hold up.

The incumbents' counterattack is formidable, too. Salesforce launched its own AI agent platform, Agentforce; Notion integrated an AI assistant as a core feature; HubSpot began bundling AI-powered automation into its existing subscriptions. These companies are rebuilding themselves around absorbing agents into the platform. The speed at which AI-native startups take over SaaS territory is now racing against the speed at which incumbent SaaS companies absorb AI capabilities. Which side wins remains an open question.

Nor can we skip the question of how mature agents actually are. Today's AI agents handle simple repetitive tasks well, but they remain limited in reliably understanding complex business logic or an organization's idiosyncratic context. Declaring "we manage sales with agents alone, no Salesforce" gets attention—but how many teams actually run that way reliably is a different question entirely. Most published cases are small pilots, or applications to simplified workflows. It's worth remembering that the rapid spread of SaaSpocalypse discourse owes as much to a mix of hope and fear as to actual operating experience. Replace your SaaS stack in a hurry without seeing this gap, and bigger operational problems—feature gaps and broken data continuity—may be waiting on the other side.

The Job Right Now Isn't Swapping Tools—It's Writing a Task List

So what does this debate actually ask of Korea's solo entrepreneurs and small-team founders?

Research on the future of work has confirmed one pattern repeatedly over the years: automation tends to replace specific tasks before it eliminates whole jobs. Researchers in the field say the real impact of automation only becomes visible when you decompose a job into a set of tasks rather than treating it as a bundle. AI agents don't stray far from this pattern. They won't replace all of Notion at once; specific tasks inside Notion—recurring meeting summaries, document drafts in a fixed format, tag classification—move to agents first. This lens sets a clear starting point for re-examining your SaaS stack: before asking whether to replace a tool wholesale, ask which tasks inside that tool could move to an agent.

The practical audit goes in this order. List the SaaS tools you currently use, and next to each one, write down the tasks you actually perform in it repeatedly. If "tasks an AI agent could handle without much difference in outcome" account for more than half of your time in a given tool, that tool is a candidate for medium-term replacement. For tools with deep accumulations of team collaboration history, customer data, and years of work records, waiting for built-in AI features is more realistic than an immediate swap. And before fully adopting any new agent tool, the right sequence is to apply it to a single pilot project for six weeks, verify the switching costs and data portability firsthand, and decide only then.

Korea's market particularities matter as well. When collaborating with domestic companies, local tools like KakaoWork and Naver Works—Korea's homegrown workplace platforms—are often the de facto standard. Even if you replace your global SaaS stack with agents, you will likely need to keep domestic collaboration tools for ecosystem reasons for the foreseeable future. The SaaSpocalypse debate is aimed mostly at the US enterprise market, and its starting premises differ from the tool decisions facing small Korean businesses. Before importing a debate raging in America wholesale, check the ecosystem of the market you actually operate in.

Here is how I would put it. What Korean founders need from this debate is not the resolve to replace their stack in a hurry. It is an accurate picture of which tasks their current tools are used for, and how much. Swapping tools without a task list is like buying moving boxes before you know what you need to pack.

Teams that merely watch the SaaSpocalypse from the sidelines will, at some point, pay the price of switching a step too late. Teams that swap their stack in a fear-driven rush pay even steeper tuition. Somewhere between the two lies the work worth doing right now.