In the past decade, investors worried that software companies would “take over the world.” With the rise of cloud services, those concerns only intensified. Companies invested billions of dollars in work automation, and businesses paid for tools that reduced the time spent on document processing. The enthusiasm was evident, and the market grew 20-fold, from $9.2 billion in 2010 to nearly $200 billion by 2023.
Then AI arrived. Now, it threatens to erase much of what SaaS companies spent years building. AI agents introduced a new model of work with a single interface that undermines the traditional SaaS business model. Earlier, these companies suggested users separate systems for each business function. We had CRM for sales, ERP for operations, and BI tools for analytics. Users needed to understand processes and know exactly which system they were working in. Agentic AI made processes much simpler, and today, they can enter a single prompt to receive instant results.
Why a “single interface” is so dangerous
In January, one of the leading AI model developers, Anthropic, introduced a Claude Cowork plugin. This system may operate autonomously across business functions and solve work tasks. The user provides Claude with some assignment, for example, “prepare a contract for this deal,” and the system executes the entire chain of actions using data from the workspace. So, it replaced the entire workflow in CRM or ERP with a single interface.
Investors quickly caught what this could mean. On the day the tools were released, the market capitalization of SaaS companies dropped by $300 billion. Now, many companies are actively integrating agentic AI into their operations. Klarna, with 118 million customers, moved away from SaaS solutions to its own AI infrastructure. According to the company, within the first month, the new system handled 2.3 million requests and generated $40 million in revenue.
One could argue that Klarna is a fintech company which must adopt new technologies quickly to survive, nevertheless, even traditionally slow-moving industries recognize the value of agentic AI. Shell, for example, uses more than 100 AI applications at different stages of oil field development. JP Morgan is deploying tools to write performance reviews. Even in healthcare, the Mayo Clinic made it possible for doctors to receive automatically generated treatment plans. AI is significantly simplifying interactions with enterprise systems and making them far more accessible.
Will SaaS lose all of its money?
As I mentioned earlier, the shift in how interfaces are used disrupts traditional SaaS business models. They used to generate revenue just by expanding their user base, yet, they didn’t care who exactly used the system or how intensively. Large enterprises with huge teams were the main revenue drivers, and they paid for access just in case. Teams may rarely use functions, and in that sense, SaaS companies are similar to gyms, where the best customers are those who pay but rarely show up. It was quite easy to scale and predict such a model.
However, businesses no longer need to pay for hundreds of SaaS seats because AI agents can replace them. Klarna, for example, stated that its AI agent may replace the work of 700 employees. It’s not difficult to imagine the scale of labor (and subscription numbers) that could be erased as it scales to large enterprises.
This shift also erodes interaction with the product. Companies were paying SaaS companies because they see employees working in CRM or ERP systems, performing visible tasks. There was a clear sense of where the money went. If your interaction is reduced to entering prompts in a single interface, the business may start asking pricing questions.
All of this puts additional pressure on SaaS companies. Claude Cowork may not replace Salesforce today, but its presence definitely changes customer behavior. Clients could refuse to pay for new services and rely on external agents instead. As a result, the growth of SaaS companies depends on the environment and customers rather than on their own efforts. And we can see large SaaS companies already trying to adapt their business models, because they can’t rely only on user growth. Salesforce, for example, is promoting both models on an action-base and fixed subscription with unlimited access to agent functions, while ServiceNow is introducing pricing tied directly to outcomes.
How to survive in the age of AI
Some may conclude that the introduction of AI will make SaaS disappear completely. Indeed, many small and mid-sized companies without unique expertise are likely to be replaced by emerging AI agents. However, enterprise software is much bigger than just products and interfaces. Large companies have access to unique business data, accumulated expertise, and accountability. If something goes wrong in a CRM system, you know who to contact and who is responsible. But who do you turn to if Claude Cowork corrupts your database or deletes critical emails?
Moreover, SaaS platforms are tightly integrated into business processes, making them difficult to replace overnight. It is also important to consider traditional small and mid-sized businesses with narrow specializations. These companies are accustomed to their existing systems and often have limited incentive to switch to new technologies.
So far, the market actually reflects this reality. Despite quick progress, AI still accounts for only about 6% of the global SaaS market. Analysts expect this share to grow to around 30% over several years, but it doesn’t change the fact that companies will need to adapt to a new environment. To prosper, SaaS businesses need to put AI into the core of processes. They can’t simply wait for surface-level integration of hyped AI functions to generate more revenue.
This shift in processes is closely tied to monetization models. Since enterprises are likely to stop paying for the number of seats, SaaS companies should rely on pricing based on outcomes or delivered services. At the same time, they may diversify their revenue and use their greatest competitive advantage they have. Years of close ties with hundreds of businesses have allowed SaaS companies to build unique expertise and collect data that can be used as a new revenue source. AlphaSense, for instance, builds its strategy around combining AI with trusted information, allowing it to remain relevant even as users migrate to agent-driven interfaces.
In the end, it shows that SaaS business is not limited to a single technology or product offered to customers. Underneath the infrastructure layer lies real knowledge, built only through years of successful work with clients. Companies need to rethink their role in existing value chains and identify the insights they can bring to their clients.
