The “SaaS-Pocalypse,” which I commented on earlier this year, is a persistent yet cresting technology and Wall Street analyst narrative that rests on three flawed assumptions. First, advanced AI assistants and autonomous agents will make traditional software-as-a-service (SaaS) enterprise applications and their business models obsolete. Second, AI will allow ordinary workers to vibe code custom software rather than buying it. And third, ignoring mounting evidence to the contrary, business software providers will not adapt to new technology by failing to embed AI assistants and agents in software. ISG Research asserts that by 2028, almost all business software providers will have augmented the capabilities of their applications with some agentic AI capabilities to lower costs, upskill users and improve customer service and agility.
The same was supposed to happen with the advent of the cloud, where "cloud first" providers were claimed to have the same sort of sweeping advantages over incumbents. The cloud was very disruptive and gave some upstarts a crack in the market to exploit, to the disadvantage of market leaders. Some (for example, Salesforce and Workday) were extremely successful, while many other start-ups either failed or achieved no more than also-ran status. Meanwhile, in the various domains of business software, most of the market leaders from 20 years ago are still around. The essential difference was that the cloud mainly required reconceiving the architecture (challenging but not impossible) rather than a totally new conception of the delivery of the software's functionality, last seen in the 1990s.
AI is causing—and will continue to cause—sweeping changes in the design and functionality of all software, but will not radically transform the contours of the business software landscape. The transposition of the cloud-native model to an AI-native narrative misses the point that artificial intelligence is both a foundational and additive technology. In the case of the latter, it takes much less effort to supercharge decades of subject matter expertise and the ability to manage the nuances that apply to specific business models than starting with an AI platform and expecting it to be quickly trained to replicate that knowledge.
Moreover, even with co-pilots and other substantial enhancements to the functionality of personal productivity software (such as spreadsheets, word processing and presentations), the dividing line between these tools and enterprise software will remain. Systems of record still matter, and their increasing use of AI and agents are built around and depend on reliable core systems of record for deep domain knowledge, security and governance.
The second part of the narrative, which expects a blossoming of do-it-yourself applications, is also conceptually faulty. DIY apps can substantially multiply individual and small workgroup efficiency and capabilities, but lack the critical functions of governance, maintainability and security. The experience most enterprises had with Lotus Notes should be a reminder of how these sorts of applications multiply and the effort required to prevent them from becoming an unchecked growth of IT kudzu. Of course, we could count on agents to do the work of curating and testing those libraries (sarcasm).
The notion that vibe coding will replace packaged software reminds me of a conversation I had with a colleague in 1994. I had just purchased a Windows-based contact management application for $19.95. He (a long-time computer hobbyist) asked me why I didn't build one myself. The answer should have been obvious.
Narrowly focused software for automating and facilitating specific individual or small-workgroup tasks will become increasingly possible and contribute to the expected productivity boom. They will not be a practical replacement for most general software categories. You could build your own, but the true cost of creating and maintaining the software—not to mention the likely gap in capabilities, governance, security and control—will limit penetration of these markets. The main exception is edge cases where an enterprise has to make major modifications—and compromises—with an existing third-party application because its business model is sufficiently different from the mainstream. These custom applications, created by internal IT staff or third-party systems integrators, will be far more affordable than before while being enterprise-grade. Yet, edge cases don't move markets. With some exceptions, the shape of today’s market structure in business software will persist, at least through the end of the decade.
It’s likely that the business software landscape will change as a result of AI, but it’s more likely that, rather than the wholesale devastation of today’s business model, individual providers will gain share by being less expensive, faster, cleverer and more reliable in using the technology to improve productivity, speed and reliability. Pricing will adapt to the market as buyer-seller models evolve in the new age.
The main danger posed by the SaaS-Pocalypse today is people believing that it’s coming. I strongly recommend that service providers and IT departments thoroughly acquaint themselves with what’s available in software applications before investing in agent-led systems to handle end-to-end value chains. There is plenty of scope for orchestrating end-to-end value-chain automation systems, but this will mainly act as the connective tissue that ties together the business applications that execute the work, rather than replacing them. I also recommend that business application providers do a better job of highlighting the capabilities and future roadmaps.
Regards,
Robert Kugel