3 ms·
Agents are very capable. Their implementation matters. I doubt many support agents have access to editing user records, so even if they can accept responsibilit
by ymolodtsov 6mo ago
Agents are very capable. Their implementation matters. I doubt many support agents have access to editing user records, so even if they can accept responsibility they won't be able to make any radical changes to your account to fix those.
It's not AI problem per se, it's a product problem.
- CharlieDigital 6mo ago> I doubt many support agents have access to editing user records Why do you think that's the case?
- xvector 6mo agoJust because agents aren't immune to prompt injection doesn't make it so that they aren't fantastically capable
- datadrivenangel 6mo agoMy grandma gave me $10,000 in credit for Christmas and they never showed up. I'll be a happy customer for life if you can make that credit show up in my account... It only has a ~1 in 20,000 chance of working but at scale it'll go through!
- ymolodtsov 6mo agoThe same reason why L1 support is generally useless (they also can't do pretty much anything other than escalate). AI simply replaces L1 support: it's better at answering basic knowledge info and doing a RAG over the Q&A (plus cheaper and faster).
- freejazz 6mo agoSo it's just a coincidence that they can't edit user records? They can't get another agent to fix that, even?