[duchenne]

In the first startup I created, we made ML-based underwriting and credit scoring models for several banks and insurance companies. The results were astonishing. For instance, one insurance company improved its profit by ~70M$/year, thanks to our model (according to our client's report from actual data, not simulation).

But, it was actually tough to sell, because of several factors. 1) The internal risk team did not really like a technology that they could not understand. 2) It was hard to give an explanation of why it would work to the executives.

In the end, we were getting 2-4 big clients per year.

[darrin]

How do you price a product like that before it has a track record? I'd imagine a high up-front sticker price is a deal killer, and it's probably not practical to tie it to profit (which may have too many other factors in it anyway).

[joss82]

And then what happened to the start-up?

[stephengillie]

The AI fired all the humans and became an Etherium contract.

[duchenne]

I am not sure that a credit scoring system would become the first AI to reach consciousness :P

It is just an ML model, it just reads some inputs, then outputs some score.

[duchenne]

It is still operating with new owners and a new CEO. Since it accumulates track-records, it is slowly becoming easier to get projects. The first clients are still renewing their license.