Thanks for making this public. But, I would like to point out as an ML engineer that Gen AI has nothing to do with credit underwriting. It is tabular data and they probably use mostly Decision Trees and maybe some Neural Networks. My point is that it is totally wrong to assume that everything is solved by Gen AI today. Nope. Not even applicable or even if it can be “forced” to the problem it will not be that efficient from any aspect (speed, resource usage, etc.).
“we also haven’t seen NU operate for a prolonged period in a very difficult macro market. “ - not true, they born out of a recession in Brazil, from 2014 onwards. They also mentioned several times that their risk evaluation models were developed in a much severe recessionary environment.
Thanks for making this public. But, I would like to point out as an ML engineer that Gen AI has nothing to do with credit underwriting. It is tabular data and they probably use mostly Decision Trees and maybe some Neural Networks. My point is that it is totally wrong to assume that everything is solved by Gen AI today. Nope. Not even applicable or even if it can be “forced” to the problem it will not be that efficient from any aspect (speed, resource usage, etc.).
Thanks Kristof
“we also haven’t seen NU operate for a prolonged period in a very difficult macro market. “ - not true, they born out of a recession in Brazil, from 2014 onwards. They also mentioned several times that their risk evaluation models were developed in a much severe recessionary environment.
US Mexico remittance market of $700B seems a lot, from the web it seems around $60-70B / year.
Sorry Kristof. That’s a typo. Meant to be $70B. Will edit thanks for catching that.
At the same time they apply Gen AI for internal use cases - more efficient search, coding assistant, plus external use cases like customer service.