The narrative that agentic AI spells the demise of B2B SaaS is a misunderstanding of evolution for extinction. We aren’t witnessing the end of SaaS; we are witnessing the collapse of the rigid workflow, the era where software as a service was a straightjacket. We are witnessing the emergence of mass customization at scale with control.
A history of mass customization
What are good industry examples of software that allowed mass customization? They were not services, they were tools. Take VBScript. It allowed millions of micro-workflows to be coded at relative ease and it was prevalent in all businesses. Take Lotus Domino. It allowed millions of scripted agents that ran major business process orchestration for many major companies. They were all tools (Domino was a very complex one). Tools that allowed for intricacies of business processes to be coded by “in house” programmers that fit the needs of the business to a tee. Even today it’s been hard to “take them out”.
The TCO of the artifacts generated by these tools was enormous. Millions of scripts running all across an enterprise. Developers who developed them are no longer around. Broken MS Access links from VBScript programs, outdated Domino databases with massive schema maintenance problems. IT costs, risk levels, and un-manageability rose to alarming levels.
In came SaaS. Pioneered at scale by the famed CRM application we all know. It offered a “service” at a predictable cost and lower TCO, lower risk overall. A service by definition can do a prescribed set of things at scale. SaaS services support the common use case workflows that are common across industries and verticals. That is where the development and consumption ROI lies.
IT departments, CIOs, CFOs adopted SaaS at large scale for the last 25 years. It’s a great model. Fortunes were made, whole careers were made. But what happened to the unique needs of an enterprise that fell between the “crevices” of the services offered? Informal human workflows started using spreadsheets, email, wikis, phone calls, texts, slack, etc. to codify the unique needs of a bespoke business process that uses SaaS as “big blocks” of standard function available.
Generative AI: New power, old problem
Enters Generative AI and agents that have natural language interfaces. Suddenly it is possible to use agents, and yes millions of them, to operate on your enterprise data in conjunction with RAGs, Graphs, and a secure LLM, to create your nirvana customized business process and get rid of the “SaaS bloat”. Enterprise IT rises from the ashes, or better yet, all employees are IT employees similar to all employees being able to use a spreadsheet. One need not write a program to generate an agent, just “talk to create”.
Not so fast. The “millions of agents” has the same TCO and risk problem like the VBScript and Domino era. Companies are discovering that going back to a super-intelligent VBScript model does not alleviate the TCO and risk challenges. I believe therein lies the massive opportunity for B2B SaaS to redefine itself as Agentic and go well beyond codified workflows and truly offer manageable mass customization at scale.
From workflow encoding to intent-based architecture
Traditional SaaS is built on a series of decision statements. Product managers spend years guessing the most common user journeys and hard-coding them into a UI. This creates a “lowest common denominator” experience where 80% of users use only 20% of the features. There is some configurability and sometimes an “apps” ecosystem for specilization, but the result is still somewhat configurable workflows.
Agentic SaaS replaces menus and buttons with conversational intent. Instead of a user navigating through five screens to “Generate a Pipeline Report,” they express a goal.
The above results in mass customization. An agent can stitch together micro-services on the fly, every workflow becomes a “segment of one.” The software doesn’t force you into its process; it assembles a process around your specific, momentary need. A conversational interface flows naturally, accessing RAGs, tools, and LLMs as needed to progress the conversation in context. It’s not a hardcoded workflow.
The Anchor: Preserving governance and scale
While the interface becomes fluid, the underlying value proposition of SaaS remains grounded in the “Industrial Virtues” of the enterprise. The move to agents does not negate the need for centralized governance, security, and manageability at scale. In fact, the “SaaS” layer becomes the essential stabilizer for agentic chaos. It provides the systems of record and action that ensures data integrity, the permissioning engines that prevent agents from overstepping their bounds, and the economies of scale that allow an enterprise to deploy 10,000 unique agentic workflows without 1,000 unique security risks. The platform remains the guardian of cost-efficiency, centralizing the “plumbing” so that the “intelligence” can be safely distributed.
The rise of “just-in-time” UI
In the current paradigm, the User Interface (UI) is essentially persistent widgets and dashboards. In an Agentic world, the UI becomes ephemeral.
We evolve toward Generative interfaces. If you ask an agent to analyze your Q1 pipeline gap, the AI shouldn’t just point you to a slide. It should generate a custom table or chart that highlights exactly what you asked for and then disappear. We move from “SaaS as a Destination” to “SaaS as a Service Layer” that lives where the user already is (Slack, email, or a browser overlay).
The end of the “feature war”
For a decade, SaaS companies competed on feature parity. In an Agentic era, features matter less than reasoning and connectivity. Integration is at its core. The most successful Agentic SaaS will be the one that connects most seamlessly with other agents. If your Marketing SaaS agent can’t talk to your Sales SaaS agent to explain why lead velocity dropped, it remains a silo.
Agentic AI isn’t killing SaaS; it’s liberating it from the hardcoded UI. The winners will be the platforms that act as the secure, intelligent backbone for an infinite variety of user-defined workflows, and just-in-time UI, maintaining the control the enterprise requires while delivering the flexibility the user demands.
See this evolution in practice
Showpad is building AI-native enablement that adapts to your sellers, your buyers, and your business processes in real time.
