SaaS или Services-as-Software: почему ИИ — попутный ветер, а не апокалипсис для софта

ZDNet published an analysis claiming that the so-called "SaaS apocalypse" caused by artificial intelligence is greatly exaggerated, and that AI is becoming not a gravedigger for software companies, but a powerful tailwind. The piece lays out five principles for adapting to the new software market model — Services-as-Software (SaS).

What Is Services-as-Software

Services-as-Software (SaS) is a model in which AI agents perform the work itself, rather than simply helping a human do it. In classic SaaS (Software-as-a-Service), a company sells a tool, while the result is produced by the client's employees: a CRM doesn't manage deals for you, it only helps you manage them. In the SaS model, the client buys an already-completed service — legal analysis, application processing, technical support — which under the hood is performed not by a human, but by an autonomous agent.

It is exactly this shift from "selling access" to "selling outcomes" that ZDNet calls the main change in the software market. Example: a law firm used to buy document-handling software and staffed it with lawyers, whereas in the SaS model an agent prepares the document itself — and the client pays for the finished result. This blurs the line between "software" and "service," and labor markets like support and accounting turn into a market for software products.

Will AI Kill the SaaS Market

AI is not destroying software companies — it is expanding their market. That is the central thesis of the column. The ZDNet author believes that fear of a "SaaS apocalypse" rests on a false premise: that neural networks will quickly displace finished products. In reality, mature SaaS companies have something AI startups built from scratch don't: accumulated customer data, distribution channels, market trust, and deep embeddedness in workflows.

Bare "AI wrappers" without proprietary data or a customer base are easily copied and have no moat, whereas the real barrier — proprietary data and embeddedness in processes — already belongs to existing SaaS vendors. That's why, by adding agents on top of a mature product, many of them are increasing their average check, not losing it.

"AI is a massive tailwind for software companies," the ZDNet columnist states.

What Software Companies Should Do

Software companies should stop competing on the number of features and start competing on the amount of work the product performs for the client without their involvement — ZDNet groups its five pieces of advice around this idea. The logic of the SaS model shifts the point of value: what matters is not how many buttons are in the interface, but how much routine the system handles on its own. Hence the typical steps of the transition — moving pricing to outcome-based fees, embedding agents directly into the existing product, and turning accumulated data into a hard-to-copy advantage.

The practical takeaway is to measure success not by the number of licenses, but by the volume of tasks completed. A company that sells "outcome" rather than "access" also changes its metrics: what comes to the forefront is not the number of subscribers, but the share of work performed by agents, and the retention of clients who care about the outcome itself.

What This Means

The debate isn't about whether software survives as an industry, but about who captures the margin from automation: tool developers, or those who learn to sell a finished outcome. The takeaway for any SaaS team is practical — competitive advantage is shifting from a feature set to embedded AI agents that perform the work in full, and the winners will be those who rebuild their product and pricing around this logic before their competitors do.