Back in October, I wrote a post titled Selling Software vs. Selling Work, in which I noted that AI may begin to disrupt the traditional SaaS business model. As a reminder, the traditional SaaS business model allows software companies to sell their software as an ongoing service with annual recurring payments. Typically, the software is billed on a "per-seat" or "per-employee" basis. Pioneered by companies like Salesforce, this model has been enormously lucrative for large SaaS companies because they can grow as they sign up new logos, but they can also grow organically and naturally as their customers hire more employees (more seats). This has been a boom for the SaaS industry as most customers have grown headcount substantially over the last 10 to 15 years.

However, a large part of AI's promise is that it will likely eliminate a huge number of white-collar jobs, such that productive companies might start materially reducing headcount in the aggregate, posing a fundamental challenge to the per-seat SaaS model.

This is starting to become a lot more real. A couple of weeks ago, at their annual Dreamforce conference, Salesforce's CEO Mark Benioff announced a major pivot, at least in their AI product strategy. They are moving away from a per-seat model to a per-conversation model where the company will charge $2 for each customer service or sales conversation held by one of their AI assistants. This shift, in theory, will allow them to continue to thrive as companies begin cutting headcount and amping up investments in AI. For those who have read Clayton Christensen's Innovator's Dilemma, this is a pretty big deal and a brave move for Salesforce.

From the Innovator’s Dilemma:

"In essence, the dilemma is that successful companies often focus on improving existing products or services to meet the demands of their most profitable customers. This focus leads them to overlook or dismiss disruptive innovations, which typically start as lower-quality or niche products but eventually improve and take over the market. By the time the established company recognizes the threat, it is often too late to adapt."

Benioff must have read the book.

More broadly, as we consider an industry shift in this direction, it raises several important questions:

1/ Does it work? Can the AI fully replace a human or is it more of an assistant? Surely, for basic tasks, it will, but how far up the stack of human intelligence will this go?

2/ Given the low cost of supplying this kind of AI once it's built, will customers be willing to pay a price that preserves the revenue and profits generated from the per-seat model, or will SaaS companies take a significant hit? In many ways, this will come down to competition and the proprietary nature of the software. Companies like Salesforce have thrived via their scale and the fact that there wasn't a better place to go. Is AI a real, novel, differentiated technology that can't be recreated, or will it be easy to copy and will margins quickly contract?

3/There's an old saying in software that you can turn any task that you do in Microsoft Excel into a SaaS company (expense reports, project management, budgeting, sales pipeline management, etc.). What's the corollary for AI? Will thousands of point solution AI companies be built around LLMs?

4/ What does this do to white-collar employment? We've seen over the course of history that new technologies disrupt employment in the short term. Though in the long term, the shifts from the agriculture age to the industrial age to the services age to the information age, have led to long term increases in wages and employment. Is it different this time? Will AI get so high on the totem pole of human intelligence that the work humans do is materially reduced? A nice proxy for this might also be the spreadsheet. Prior to the invention of spreadsheet software (VisiCalc in 1979), there were large buildings full of white-collar analysts doing the calculator work that Excel does for us today. That shift only created more analyst jobs because high-quality analysis could be done more cheaply. When something gets cheaper, we typically do more of it.

Just when I thought b2b software was getting kind of stale…here we go again…it seems AI is going to be driving a new platform shift, and it's impossible to know where it all lands. But it's great to see companies like Salesforce protecting their shareholders by resisting the innovator's dilemma and getting out in front of what could be a massive new chapter for b2b tech.