Bridget McCormack

But as you know, in most intermediate appellate courts, there's a lot of opinions that are not written.

And in trial courts, very often there's no written opinion.

So I always want to ask, as opposed to what?

Like, if you believe that the current human-led process

overburdened justice system is one where people have lots of confidence, I want to introduce you to some folks who might disagree.

Yeah, I mean, there is that.

So if you think that's a system that instills confidence, like maybe, but I'm not sure everybody would agree with that.

Yeah, huge issue, obviously.

Yeah.

We, you know, if you were to just take your dispute, you know, all the documents in some dispute you're having and throw it into ChatGPT or into Claude, you'd get, you could get a, you could get a result right now, right?

Like anybody could do that.

It might be okay.

It might in some cases not make a mistake.

As you know, hallucinations are, you don't know, you don't know when exactly or why the frontier models are making mistakes.

But that's not what we built, and that's why we're moving so narrowly and so slowly.

Your system has to be governed, trained in, and grounded in the kind of reasoning that you're asking it to do.

So it sounds extremely narrow that we're doing documents-only construction cases, but there's a reason for that.

That's where we could build a governed and harnessed agentic system.

We keep a human in the loop to make sure that before an award issues, there were no hallucinations, even with our governance.

And we're going to be very transparent about all of our audits.