@levie: Great post on how to think about open source AI and the applied AI layer. Two things will always be true in AI. Frontie…

Great post on how to think about open source AI and the applied AI layer. Two things will always be true in AI. Frontier intelligence will likely remain at the forefront of solving brand new used cases and often be used for orchestrating and planning of any type of complex workflow. At the same time, as use-cases become mature and predictable in an enterprise context, you can begin to peel off some set of the tokens to either lower cost open or closed models *or* models that are trained for the task at hand. Doing this too early in the adoption curve of any new use-case doesn’t make sense as you don’t know what you’re optimizing for, which is why there will always be a bit of a lag here. This process can essentially run on forever as there is no end for both the benefits of frontier intelligence or tuned models. This is why the spend and token volume for both approaches will continue to go up for years to come as we’re still early on both. Again, the only reason this dynamic is possible though is because of the applied AI layer that can effectively eval their workflows in a specific domain, choose a mixture of models to solve tasks, and have enough scale to eventually train their own models for their purposes.