SaaS isn’t dead: How AI agents rewrite the SaaS playbook
Updated October 18th, 2025.
AI agents are transforming how we build, buy, and price software. For SaaS companies, this is the start of a new era.
Over the past months, many perspectives on the future of SaaS have been making the rounds in the media. For instance, earlier this year, in a podcast appearance, Microsoft CEO Satya Nadella predicted the death of SaaS and traditional software applications as we know them. His comments ignited a wave of debate across the tech world.
In this blog post, we compare three viewpoints from recent articles on the future - or death - of SaaS.
Background: The rise of SaaS
At the turn of the millennium, the concept of “software as a service” was radical. In 1999, Salesforce pioneered a model where businesses subscribed to software hosted in the cloud, rather than installing it on their own servers. The benefits — lower upfront costs, easier updates, and scalability — were irresistible.
By the mid-2000s, SaaS adoption accelerated across SMBs and enterprises. APIs unlocked integrations, vertical SaaS emerged for niche industries, and subscription-based pricing became the default for software companies. For two decades, the Saas business model defined how we bought and used software.
But just as SaaS displaced on-premises software for many organisations, a new paradigm is now emerging, powered by AI agents. Capable of orchestrating tasks across multiple systems, making decisions, and adapting in real time, these agents are changing the way we think about software itself.
#1 AI as an enhancer, not a replacement
Current headlines may sound dramatic, but perhaps AI isn’t “killing” SaaS; rather, it's rewiring it.
Bain & Company notes in the article The Great Debate: Will Agentic AI Kill Saas?, “Technological revolutions are rarely binary… Transitions lead not to extinction but to transformation, adaptation, and coexistence.”
Much like mainframes persisted alongside client/server computing, and hybrid cloud became the norm, the SaaS model will not vanish overnight.
But it will be reshaped in profound ways, and the disruptive potential of AI agents places a heavy responsibility on SaaS vendors to prove their relevance in an AI-first world.
Where traditional SaaS platforms focus on CRUD operations (create, read, update, delete) and predefined workflows, AI adds capabilities like:
Understanding natural language commands.
Predicting user needs.
Automating multi-step processes across tools.
But is this too optimistic a view?
In an interview with podcaster Dwarkesh Patel on October 17th, 2025, AI researcher Andrej Karpathy stated that agentic AI is years away from matching industry hype. He believes that the current euphoria surrounding agent AI is premature, and he estimates it will take around ten years for the field to solve the problems.
Karpathy believes that current agent-based AI systems face clear limitations in areas such as reasoning, handling multiple types of input, memory, and reliably carrying out complicated tasks.
#2 From seats to adaptive AI agents
The Forbes article Bye SaaS, We Have Entered the Agentic Platform Companies Era captures the shift well: “The dashboard-and-seat-license model is being eclipsed by AI agents: autonomous, adaptive systems that learn and execute without constant human input.”
Where SaaS applications once relied on fixed workflows and human-triggered actions, AI agents now offer:
AI-driven adaptability — learning from user behaviour and adjusting workflows in real-time.
Natural language interaction — allowing users to “talk to” software rather than click through menus.
Cross-platform orchestration — where AI executes tasks across multiple systems without manual intervention.
This evolution demands more than bolting AI features onto existing products.
As Forbes warns, mid-market SaaS companies must reinvent themselves: design AI-first products, pivot toward usage- or outcome-based billing, cut underperforming features, and invest heavily in AI capabilities — or risk irrelevance.
New SaaS business models on the horizon?
The SaaS subscription model, fixed, per-seat, was the engine of the last two decades.
In many of today’s SaaS offerings, especially those enhanced by AI, automation, or a mix of both, the value customers receive isn’t necessarily tied to the number of users actively logging in.
As a result, pricing models that focus on platform-wide fees rather than per-user charges are gaining traction, particularly among AI-first companies. Take Clay, for instance, its automation platform for lead enrichment and prospecting uses a flat, enterprise-level subscription model that stays the same no matter how many employees access the system.
AI also enables models tied to outcomes rather than licenses.
Token-based consumption (where you pay for AI processing units).
Outcome-based pricing (where fees align with business results, not usage hours).
Dynamic scaling (where AI automates cost optimisation for customers).
For B2B buyers, this could mean more cost efficiency. For SaaS vendors, outcome-based pricing sounds ideal but is hard to pull off because it only works when results clearly reflect product value, show up quickly, are valued consistently by customers, and can be verified automatically.
#3 SaaS-to-AI journey: 4 major challenges
AlixPartners offers a grounded perspective in their article Farewell SaaS, AI is the Future of Enterprise Software.
Companies undertaking the SaaS-to-AI journey face 4 major challenges that must be met if the companies are to deliver value for customers and investors:
Competition from AI-native players — leaner, faster-moving entrants can undercut incumbents on price and speed.
Increased compute and engineering costs — GenAI and AI agents demand higher infrastructure and developer investments.
Revenue unpredictability — outcome-based models tie earnings to variables outside a vendor’s control.
Operational disruptions — embedding AI means retraining staff, reengineering processes, and modernising infrastructure.
In short, the AI era is rich in opportunity, but only for those willing to tackle the operational and strategic hurdles head-on.
AI rewires the future of SaaS?
The next chapter isn’t “SaaS vs. AI.” It’s SaaS with AI at the core, software that is dynamic, adaptive, and capable of autonomous execution.
For CMOs and CROs in SaaS, technology, and IT:
Expect customer expectations to shift toward conversational, outcome-driven experiences.
Anticipate more fluid pricing and contract structures tied to measurable business value.
Invest in AI capabilities now, not as bolt-ons, but as foundational elements of your product and go-to-market strategy.
