E185 - AI - Beyond the Context Window - Johan Thulin

But for your

own purposes or was it for some customer or what was the purpose of trying to do data augmentation at that time?

To solve the lack of data in Europe after the EU Data Act, we didn't have a single legal data set really.

Every data set that's out there is polluted in some kind of way.

But was it for some project, some company, some customer? For what purpose were you trying to find data at that time?

For the EU to be able to continue to develop.

To have some open source data sets that you can.

Exactly.

I see.

Okay.

And that was the original purpose of FIGO to do that, right?

Yes.

We saw there was a lot of students sitting on data.

We figured out what if we can get use of that data.

We know how to build data sets and data set pipelines.

We know how to argument data.

But there's big issues with data argumentation today.

We are still using manual annotations, basically sweatshops that sit and annotate data.

But we started to say that, no, we could probably train models and do the annotations for us and do it in an agentic way.

And then we were starting to look at agentic workflows and orchestration.

And we realized that, oh, we could probably need a really good memory in our agent orchestration, in our pipeline.

And also just to help us with our company.

So I...

Went back and took up my old research again.

Started building our own memory for a pipeline.