For the last twenty years, the Software-as-a-Service (SaaS) playbook was simple: identify a business pain point, build a cloud-based tool to solve it, and charge a monthly fee per user (“seat”). This formula generated some of the highest valuations and most consistent growth in economic history.
But in late 2022, the ground shifted. Generative AI arrived, and suddenly, the traditional SaaS value proposition started to look shaky.
A wave of panic rippled through the tech world. Stocks took a quick dip on headlines of AI advancements and quickly began positing, “Is SaaS dead?” Predictions started that small, AI-powered startups would dismantle multi-billion-dollar incumbents. The rise of sophisticated Large Language Models (LLMs) changed the equation. If an AI can write code, analyze data, and create content autonomously, do companies still need massive, off-the-shelf software platforms?
No, the SaaS industry is not dead. But the era of passive “systems of record” is over.
SaaS is facing a Darwinian moment. It is not an extinction event, but an urgent evolutionary imperative. For SaaS firms to survive and thrive in the future, they must move beyond simply providing software as a tool and start providing software as a solution — powered and delivered by AI.
Where the “SaaS is Dead” Narrative Comes From
The anxiety surrounding the traditional SaaS model stems from several genuine disruptors introduced by AI:
The classic SaaS model relies on human users interacting with software to get work done. A CRM exists for salespeople to input data and send emails. An HR tool exists for people managers to process reviews. If AI starts automating the everyday tasks within the platform, the need for dozens of expensive user seats diminishes rapidly.
Historically, building enterprise software required significant capital and engineering talent. Today, AI helps developers write code faster and better. It is now possible for a two-person team to build a complex application in months that previously required a team of fifty. There are still other barriers — marketing, distribution, data — but a central one is diminished.
Many SaaS products that rushed to add AI are simply “wrappers” — minor UI interfaces that send prompts to OpenAI’s API. The problem? The underlying AI provider can easily add that functionality themselves, rendering the SaaS product redundant.
What Must Happen Next
For SaaS firms to provide defensible value, they cannot just add AI features; they must become AI-native. This requires moving from being a “System of Record” to a “System of Action and Insight.”
The next generation of SaaS — call it SaaS 2.0 — focuses less on the user interface and more on the outcome. In the old model, you paid HubSpot so your employees could manage marketing. In the new model, the SaaS product itself will autonomously execute marketing campaigns, optimize them in real-time, and report the results. The human user shifts from being the “doer” to the “editor” or “curator.”
How SaaS Firms Can Adapt
There is a defined path forward. It requires leverage, integration, and a willingness to cannibalize — or redefine — one’s own old business models.
Generic LLMs like ChatGPT are smart, but they do not know a company’s unique internal context. An HR SaaS company with ten years of anonymized performance data, compliant workflows, and company-specific policies has a massive “data moat.” By fine-tuning AI models on this proprietary, clean, and structured data, SaaS firms can offer accurate, bespoke insights that a generic AI cannot replicate. Data is the new defensibility.
The most vulnerable SaaS products are horizontal — tools that apply broadly to many industries, like basic project management. The safest are vertical — those built specifically for a niche, such as dental practice management, construction logistics, or insurance claims processing. Vertical SaaS platforms integrate deeply with specific industry regulatory environments, unique legacy systems, and complex internal workflows that AI can’t easily navigate without substantial, pre-existing domain expertise.
Integration and orchestration are key. If a SaaS product is the “brain” for a specific business function, it must have “hands” to get things done. SaaS firms must build AI agents capable of autonomous or supervised multi-step tasks. An AI in a project management tool should be able to read a project brief, automatically assign tasks to the correct team members, set deadlines based on historical velocity, and flag potential bottlenecks before they happen. This is a system of action.
This is perhaps the key adaptation. The user-seat model is fundamentally misaligned with an AI future where fewer human users are needed per application. SaaS firms must move toward consumption-based or outcome-based pricing.
- A flat integration fee for access to the core platform
- A usage fee based on computational consumption (AI tokens, pages, or API calls)
- A value fee for results — e.g., $10 per qualified lead generated, or a percentage of cost savings identified by the AI
This shifts the metric from adoption (how many people have a login) to value (what did the software actually achieve).
Software as a medium for business operations is not going away. What is dying is the tolerance for software that requires the user to do all the heavy lifting.
The SaaS firms that will fail are those that ignore AI, move too slowly, avoid experimentation, or treat it merely as a minor change. The firms that will win are those that leverage their existing data moats, re-architect their platforms to act more autonomously — with the right elements of human oversight — and realign their pricing to the actual value delivered.
The AI evolution is not a closing door; it is an opened gate. The SaaS industry is simply walking through it into a new, more automated, and far more powerful era.
More thoughts to come.
This is part of an ongoing series on technology, business strategy, and the world being reshaped by AI.
Return to alancyates.com