AI Investor and CEO Rob May Discusses the Value of the Post-Model World Thesis

AI models continue to improve, but according to Rob May, those improvements yield diminishing returns relative to what the systems they support can provide. Evaluating the usefulness of various AI models to inform which ones to adopt served as the foundation of the decision-making process for many businesses and organizations, in part because improvements to those models were presented in ways that invited comparison.AI experts like the investor and CEO Rob May have since found that this approach to evaluation is misleading, however, prompting May to pursue another direction by founding a company based on what he calls the “Post-Model World thesis.” Instead of emphasizing models, this new AI investing framework evaluates the systems in which they operate. May describes, “the model is the engine, but customers buy the car,” drawing on this and other analogies in several writings, including his upcoming book and his Substack “Investing in AI.” Putting Theory Into Practice May’s latest contribution to the field of AI and AI development is his Post-Model World thesis. This theory suggests that AI models, while still important, are losing value to the broader systems they power. May writes in his online newsletter Investing in AI that “[models] are becoming the foundation, not the differentiator,” suggesting that the focus of value creation has shifted from the models themselves to the systems that can use them best. Part of this process involves what May calls “model fragmentation”: the idea that systems work best when using several smaller systems rather than a single large, potentially inflexible one. To test this theory of model fragmentation, May founded Neurometric, an automated token-engineering platform designed to help users intelligently allocate their workload. This approach examines whether using multiple small models is more efficient than relying on a single large model. As the company works on updating processes such as MCP protocol standardization and model routing, May comments on their progress and developments in the latest manuscript of his upcoming book, giving readers a more complete picture of the current state of AI investing. Testing AI’s Limits and Capacities In addition to testing the limits of model fragmentation through Neurometric, May has also demonstrated some of AI’s current capabilities through his writing. He states that 90% of his upcoming book is human-written, with the remaining 10% coming from a chapter marked as AI-written to show readers the current state of writing tech with LLMs. This willingness to test as he goes is what has helped May stay up to date in what is often described as a whirlwind industry. With an angel and VC track record spanning over 75 AI companies, he grounds many of his decisions in real-time theorizing, regarding both his investments and his approach to leading Neurometric. Today, May continues to refine his theories on model fragmentation through both his writing and his work with Neurometric. He has also signaled plans to continue publishing as both an operator and an investor, and to deepen his public markets commentary alongside the book. In doing so, May maintains his position as someone capable not only of learning from what others have to say about AI, but also of actively contributing to the field. This story was distributed as a release by Jon Stojan under HackerNoon’s Business Blogging Program.