Mistral pairs models with applied scientists to reshape workflows

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

1. 8:27

   i think you're seeing it the most actually being at the forefront talking to a lot of these enterprises there's this big conversation around sovereignty and what it actually means to be a business have full ownership over your data over your intelligence but as we alluded to earlier sovereignty around cyber defense and having cyber defense capabilities is increasingly growing in importance so tell me a little bit about how you guys are positioned to solve that Yeah, so Mistral really was born as a frontier lab.

2. 9:00

   And a lot of the investments that we have received over the years are really deployed in continuing to deploy best-of-breed open-weight models. But we've also very much expanding north and south of that very important intelligence layer. We deploy those models now on our own infrastructure. Being sovereign, it's also enabling customers to deploy those models anywhere. We have data centers in Europe, in the Nordics, and growing everywhere.

3. 9:29

   That's the first part. The second part is what we've also learned is that, have you heard the 95% of AI projects are failing? It's also really a very important concept that for enterprises to unlock value out of AI, it's just not only about productivity of employees and developers. Of course, it matters, but it's also about completely transforming their core processes. And for that, north of the intelligence layer, Mistral has built our own really platform that does enable core workflow transformation.

4. 10:08

   And as part of that, model customization. Model customization now is a Lego bricks as part of our platform that brings our own ability of customizing models as a science organization, bringing that to the large enterprise so that interns, they can customize their own models. So really, in order for us to enable that autonomy, that sovereignty, that control, we're doing that with this vertical integration between infrastructure, model apps and developer tools.

5. 10:39

   All of that powered by a very strong forward deployment engineering arm. This is now an overused marketing term, but that team has been built over the past three years. Those are applied scientists that are enabling that model customization. An example of that, we are very strong in industrial engineering. We have great physics models. And those physics models is what is powering today really tools such as crusted testing assistant at BMW.

6. 11:10

   You cannot do that with closed source models. You need the ability to access specialized models that are open and that can be customized. So the way we unlock that value is really between that vertical integration of software and models, forward deployment engineering arm, and a very specific focus on financial services, manufacturing, oil and gas as our primary really industries, because that's where a lot of the control and customization is absolutely needed to unlock value. Yeah, I want to talk specifically