What do LLMs mean for the next generation of SaaS startups? (Especially those in professional services)

SaaS and its Discontents

To paraphrase Mark Twain, “rumors of [SaaS’s] demise are greatly exaggerated.”

Capital and subscription dollars have accumulated from Anthropic to OpenAI at the foundational application layer, which now underlies a plethora of new "AI" features integrated into existing applications. However, as capital flows toward these foundational models, profits may erode as open-source models become refined for specific applications, potentially supplanting general models. I believe the application layer, where UI and UX come into play, and the interaction of these models with "jobs-to-be-done" will be the focal point for profit creation. This suggests room for SaaS companies to disrupt incumbents and establish themselves as the new default OS systems for enterprises.

While some view these models as a panacea for workplace challenges, Large Language Models (LLMs) all share a fundamental limitation: they are inherently "artificially intelligent." Instead of genuinely understanding causality and logically deducing results, they can only identify the correct answer using probability. This marks the difference between probabilistic and deterministic thinking—they can only indicate what is likely accurate, not definitively true. If LLMs achieve this with a high degree of accuracy, some argue it's practically the same thing. Nevertheless, LLMs have yet to demonstrate this level of accuracy without supervised tuning and instructions, particularly in professional services.

Hence, these models derive value when tuned to solve specific business tasks demanded by users, such as those in law, accounting, and medicine, as they reduce Type-I errors. Consequently, a model becomes genuinely useful in the long term when it consistently surfaces and structures the data crucial to business operations.

Yet, tuning alone isn't enough. For a new business to disrupt existing ones, it must challenge the status quo and maintain a competitive advantage capable of generating economic surplus. This surplus materializes when these new companies establish a lock-in effect through data management, insights, or networks they can leverage.

The next frontier for these emerging businesses will involve exploiting the enhanced interactions facilitated by new interfaces with LLMs to create these competitive advantages. Just as GUIs supplanted the command-line interface, opening doors for new levels of human-computer interaction, we now stand at a similar technological crossroads.

Consequently, I foresee a prolonged period of disruption as both incumbents and new players leverage this technological shift to secure market positions and drive substantial productivity gains.

At Global Millennial Capital, we actively seek game-changing companies poised to emerge as the new operating systems for businesses, capable of harnessing this historical disruption. If you are building a company disrupting professional services using LLMs and have a vision for the future of your industry, feel free to reach out via email at [email protected].