AI Flywheels
The most compelling AI-first companies will find a clear flywheel: a good product begets users begets more data, easing research complexity to build stronger generalized models which begets great products.
— Suhail (@Suhail) January 28, 2023
I've written before about flywheels more generally within digital products and the building of technology and digital companies. I believe that flywheels will be even more important for AI companies going forward, as people scramble to find ways to differentiate amongst an ever-expanding sea of competition.
While it's likely that this will change as open source solutions become more sophisticated, the majority of large language model development that is happening at the moment is tied to OpenAI - the underlying technology is therefore not something that will differentiate products, but that is not to say there aren't other ways to differentiate products that rely on LLMs.
The '7 Powers' is a framework developed by Hamilton Helmer, which identifies seven ways in which a company can build and grow an enduring advantaging (i.e. a moat) in contrast to their competition. A summary of them is: