Meet The Startup Probabl by the Creators of Scikit-Learn

You know, there's a general thinking that Gen-AI requires a lot of compute. That is true, but we are not in the space of Gen-AI. We are in the space of statistical machine learning initially. Scikit-Learn, the number one framework for machine learning, exceeding 5 billion downloads to date, built at INRIA, the research center in France, adopted by everyone who was getting into data science. It is the de facto framework. On GitHub, there's 1.3 million projects that depend on Scikit-Learn. scikit-learn has been initially conceived for cpu use but now for certain algorithms estimators as we call them we are better off offloading to gpu we maximize the ability to deploy workloads at scale because you will have new generations of chips that will just naturally feed into this abstraction layer so that not we don't have to rewrite everything every time in the partnership with inception is the relationship we have and the access you give us to your colleagues so that that we can engage in this specific aspect. We are very much projecting to be the tabular AI company worldwide in five years. The open source ecosystem is very dynamic and tackles many problems because of the community aspect. Every single software engineer, data engineer, data scientist uses open source day in, day out. How do we steer that in a way that is enabling business cases? We need to maximize total cost of ownership, maximize the amount of tokens per gigawatt. In other words, the number of intelligence units per gigawatt. Our aim is to actually create a layer that is seamless to the end users. Now agents consume. We've had 2 billion downloads in the last 12 months, most of which come from agents.