And this question comes in from Mark, who's in, uh, Des Moines, Iowa, and Mark is asking, uh, and this is for you as well, Ziad, this is, um, how do you decide which customer intents should stay human only, which should be AI assisted, and which can safely move to fully agentic resolution?

Uh, that's a great question.

It comes down to how much of it can you satisfactorily respond to with AI agents, and that comes down to do you have the right context, do you have, like, the right resolution protocols? So typically what you would wanna do is you wanna get that process.

So you start with the intents that you have all the information for, that you can actually have human-level agentic responses to, and test them with your customers, see if that's satisfactory, and it's actually not dropping the quality of your interactions with your customers.

And if you see that whole, then that's, the logical thing then would be to just keep it at agentic.

And for cases where, okay, you might not have the full context, you might need some humans in the loop, then you keep it at that.

And basically it's based off of, do you have all the info that you need, and by making it agentic does that hold the quality, and makes sure that your customer experience doesn't degrade by making it agentic.

And Ayon is asking, "For a support team just starting with agents, what's the first narrow intent you'd choose so your governance engine can prove its value without risking account balances or promises?"

And this question comes in from Mark, who's in, uh, Des Moines, Iowa, and Mark is asking, uh, and this is for you as well, Ziad, this is, um: How do you decide which customer intents should stay human-only, which should be AI-assisted, and which can safely move to fully agentic resolution?

Uh, that's a great question.

It comes down to how much of it can you satisfactorily respond to with AI agents, and that comes down to do you have the right context, do you have, like, the right resolution protocols? So typically what you would wanna do is you wanna get that process, so you start with the intents that you have all the information for, that you can actually have human-level agentic responses to, and test them with your customers, see if that's satisfactory, and it's actually not dropping the quality of your interactions with your customers.

And if you see that hold, then that's...

The logical thing then would be to just keep it at agentic.

And for cases where, okay, you might not have the full context, you might need some humans in the loop, then you keep it at that, and basically it's based off of do you have all the info that you need, and by making it agentic, does that hold the quality and make sure that your customer experience doesn't degrade? By making it agentic.

So it's a matter of, like, tweaking, reviewing, and so on and so forth.

And Ayaan is asking, "For a support team just starting with agents, what's the first narrow intent you'd choose so your governance engine can prove its value without risking account balances or promises?"