The tech industry is experiencing significant disruptions, with the rise of AI agents at the forefront of this transformation. With widespread layoffs and automation altering the landscape of traditional white-collar jobs, Software as a Service (SaaS) companies must adapt their pricing models to the new paradigm. AI agents will take over many tasks that white-collar workers perform today. However, to do this effectively, they will need access to the same systems and repositories as their fellow organic co-workers. For example, AI-powered SDRs should have access to CRM systems to perform effectively. Similarly, AI-powered developers will need access to code repositories, development environments, ticketing systems, and more.
At CarriedAI, we are developing AI investment teams for institutional investors in private markets (PE, VC, VD, etc.). To provide our agents with the same data access as human workers, I am engaging in numerous conversations with private market data providers. Our goal is to develop products and pricing models suitable for agentic clients. This involves combining API access with consumption limits that resemble what a human worker would use. Although these conversations always take time, I am very excited about the progress we are making and the understanding that data providers have regarding AI-suitable models.
This blog post explores why SaaS companies will need to shift from per-seat pricing to accommodate agents accessing the platforms instead of human workers and how they can effectively implement these changes.
The Current Landscape of Tech Layoffs and Automation
As of early 2024, tech companies have laid off 438,174 employees since the start of the COVID-19 pandemic. In 2023, there were 262,242 tech layoffs across 1,186 companies. The first few months of 2024 have already seen 60,000 job cuts across 254 companies, with significant layoffs from major companies like Tesla, Amazon, Google, TikTok, Snap, and Microsoft. In January 2024 alone, 19,350 tech employees were laid off.
It is questionable whether the lost jobs in tech companies will ever be refilled or are forever gone. A Brookings Institution study found that 25% of U.S. jobs paying over $57,000 annually could be disrupted by AI. The World Economic Forum projects that AI will displace 85 million jobs across 26 countries by 2025.
The Challenge of Per-Seat Pricing
If white-collar jobs are challenged by AI then the first derivative of that is a challenge to the top-line of SaaS companies that price their products on a per seat basis. These changes underscore the need for SaaS companies to rethink their traditional pricing models.
Per-seat pricing, where companies charge based on the number of users, has long been a staple in the SaaS industry. However, the advent of AI agents is challenging this model. AI agents can perform tasks traditionally handled by humans, potentially reducing the number of human users needed. For example, according to a Gartner report, AI could reduce call center staffing by a staggering amount, leading to a significant revenue loss for SaaS companies relying on per-seat pricing, despite the continued or increased value of their software.
Alternatives to Per-Seat Pricing: Consumption-Based Pricing is not always the solution
To address these challenges, SaaS companies should consider alternative pricing models that better align with the value they provide. One suitable alternative could be the Consumption-Based Pricing (CBP), which is fairly widespread in infrastructure software companies like Snowflake or Datadog. However, this model is not suitable for every company and in some cases it introduces other problems into the mix.
The CBP model charges customers based on usage metrics, such as transactions processed or data utilized, rather than the number of users. While this in theory sounds great, in practice it becomes challenging to implement in many cases. Imagine Salesforce charging for the number of deals won or Jira for the number of issues resolved in a given month. This would generate strong push-back by their customers that would probably see these pricing schemes as unfair and intrusive.
Agentic Seat Pricing
In an era where AI agents become ubiquitous, wouldn’t it be easier if SaaS businesses offered “agentic seats”? If AI agents represent the workforce of the future, it is most natural for SaaS companies to price their products using agentic seats as the unit of pricing. However, in order to be fair for both the customer and the vendor the agentic seat pricing should fulfill the following requirements:
Granular API Access: Agents will most likely operate on SaaS platforms through APIs and therefore they will need one API key for each agent. Currently, most SaaS platforms provide one API key per customer, not per seat. This means the number of API keys that SaaS platforms need to handle will increase by a factor of 100 to 1,000 when moving from one-key-per-customer to one-key-per-agent.
Consumption Limits per Agent per Month: A set of consumption limits for each agent should be established. These limits should likely mirror the activity level of human workers more closely than the consumption levels of an API today. If a human worker can process for example 30 deals per day, the activity of the agentic equivalent through the API should be in the tens of deals, not thousands, despite the agent’s computational capabilities.
Feature-based Pricing: While not a requirement, this presents an opportunity for SaaS vendors to better align pricing with value. Similar to human workers, agents could access different sets of features through the API depending on the pricing plan that they subscribed to.
Conclusion
The rise of AI is transforming the SaaS industry, necessitating a shift from traditional per-seat pricing to more flexible models that reflect the nature of the new users. By adopting Agentic Seat Pricing or Consumption-Based Pricing, SaaS companies can better align their pricing strategies with the increasing presence of AI agents in the workforce.
