[disgruntledphd2]

Personally I think the big benefits are gonna be in pretty specific business processes which are currently done manually. Lots of KYC and data entry between systems where it wasn't plausible to automate before are now doable.

I'm pretty uncertain around the businesses of foundation models but the tech is definitely useful.

[nstart]

Yep. I like to sometimes think of Agents as a slow but tireless unofficial API glue between completely disparate elements. I dislike the approach the companies took in laundering information through them. But their ability to work through clerical work that no one should be doing ideally has been incredible for me. Of course there's nuance to that last sentence. Someone needs to have an eye on certain things in domains like finance. But it's up to companies to be smart here and ask how AI can augment that process rather than trying to get rid of it. For example, collecting all relevant context and presenting it on demand for a suspicious transaction flag would eliminate what could be days of inter departmental wrangling in a normal work process.

My only side note of sadness here is that companies are more likely to implement such stuff in a haphazard way rather than anything actually thoughtful.

[disgruntledphd2]

> My only side note of sadness here is that companies are more likely to implement such stuff in a haphazard way rather than anything actually thoughtful.

This will definitely happen, but the trend will end up towards automating a bunch of this stuff with manual checks on a subsample to ensure that it's working effectively. It's rather like moving from phone/mail based ordering to online ordering, in that it'll take a while but it's almost certainly gonna happen.