Joel Hron

Oh, 100%.

And we hear that loud and clear from our customers.

Like, you know, if we just spit out an answer, I don't care how right it is.

Like, there is very little trust of it, as there should be, I think.

I think trust is earned over the course of time.

And, you know, we need to do everything we can to earn that.

But I think in the time between now and then, like explainability and transparency are super important.

We actually just had a product review this morning with our tax team.

And we were going through the same topic about what the right levels of transparency and human agency is necessary in the process.

And it is a balance between giving the user enough of that, but also not making it so they're having to validate every step that the agent takes, because that's not really helpful either.

Kind of needs a purpose, yeah.

Yeah.

So trying to find the right balance of what's enough there is a learning experience as we sort of work with customers and clients on this type of thing.

I think the other thing that we've done in terms of explainability overall is...

really try to include the context of the source of any conclusion within the application itself.

So for instance, in a deep research answer, there's deep links to resolve to the actual cases that supported this answer.

There's flags.

I mentioned Keysight earlier.

There's like

The same UI flags that exist in our legacy application exist in this answer to give the user these visual clues about where to go next and how to launch from here.