Grant Harvey

Step two, talk about AI agents.

Step three, magic happens.

You deploy an AI agent.

How do you know if it's doing a good job?

Okay, and how do you get an evaluation?

You need to understand what the measures of a task being performed acceptably and accurately are.

And often the best way to get those is to do the task a lot.

Yes, right.

And that's what our people can do is they can actually perform this work and we can start to build up the evaluative sort of set of metrics to then deploy an agent effectively.

Because candidly, if a company is going in

and saying, I'll deploy AI agents, it'll all be tech.

They're either there saying, hey, client, you used to do a hell of a lot of work for us, or, hey, we are YOLOing this completely.

And neither of those is particularly compelling for an enterprise.

And because we price on outcomes, we price on unit basis, et cetera, we aren't saying, oh yeah, we'll bring our people in and you'll pay hourly and we'll just do this forever because it's a wonderful business model for us.

We're saying our people are going to come in and they're going to work

over the course of this project to make themselves obsolete in this case.

Or they're going to be like, actually, this piece is so subjective and so complex and so strange that actually you need people.

And that's going to be the best way to live quality because we are not there.

We're there to deliver the outcome, not the mechanism for the outcome.

We're not there to sell a certain way of doing something.