For anyone who's curious about what happens behind the scenes, what does it actually take to build an AI product? Can you walk us through the process from data and model development through to testing and deployment?

Yeah, I'm very happy to outline this a little bit.

And indeed, this has also been changing very much with generative AI and large language models.

I think traditionally building an AI or machine learning product usually started with data and maybe first before that even with a business problem, right? So if you want to build a product, hopefully you build a product for a need that your users really have.

So you still need to understand your customers, your users, what kind of functionality would help them what new product maybe is not there that you should build.

This is still the same.

But for building traditional machine learning models, so deep learning models, we've then had to go and say, okay, now we need data and maybe we need a few hundred examples, maybe a few thousand, maybe 10,000.

What's one piece of advice you'd give someone who was looking to break into data science at the moment?