So it's a really a fascinating time, but what do you think, Barr?

I think the jury is still out.

Like, I think very few companies and enterprise have actually gone through AI transformation, so trying to measure that or assess that is, is just too early.

I think there's some anecdotes, um, you know, here and there, and I also think it's impacted industries very different.

So if you look at the areas where AI has been able to make a significant impact, there's proda- primarily three industries.

The first is coding, and so impacting engineering and engineering headcount.

Um, there, pretty sure we've seen growth based on, again, sort of the, uh, you know, limited information that I have.

Mm

So I'm just curious about like, you know, when you guys are serving these like giant enterprise, are they mostly like building internal agents or external agent? And then what is the most popular use case for agents?

So good question.

I think there is probably...

I'll categorize it into three types of use cases that I see.

The first one, which almost all companies do, and I think it's because the thing about agents is it helps give access and give knowledge to people that historically didn't have that.

That's one of the powers of AI.

This type of agent, you know, you can call it like conversational analytics or ask your data or I'm trying to think if there's different names.

And how do you like make sure that when the agents are orchestrated, like it's like each step, there's some sort of like limits, but also not slowing the agent down from like, you know, fact checking everything and stuff like that.

Mm-hmm

... or an organization actually upskill.

And I don't f- I think it's very hard to understand what AI can do without actually working with it.

So I think that's the first phase.

The second phase, where we are now, is where we actually are demanding more from ourselves and from organizations, and actually looking to, to see our agent solving a real problem.

Now, the thing is, most agents, in order to make sure that they're running in production, require thinking about it as in a systems view, if you will.

And so at MontyCo, we actually found that there's four different things that can go wrong or four different, totally different layers of what goes into when an agent goes rogue.

Okay.

What are some of those misconceptions?

Great question.

I think one of the biggest m-- you know, one of the biggest, I don't know if this is a misconception, but it's more of a fear, is that, you know, AI will sort of eliminate us all.

Um, and actually, I, I, I don't believe that is the case.

I think, you know, uh, for, for the last however many years, we, you know, one for humanity, zero for technology, right? We're still here.

Um, and so, you know, I, I'm an optimist.

Uh, you know, I, I, I think, um, you know, maybe to, to, to give you sort of a more practical example, you know, one of the chief data officers that I met with this week shared with me, you know, how did, uh, stoplights begin, right? It started with, you know, someone actually holding, like, a red flag, uh, in the street and, and stopping cars, and then over time, we introduced, you know, a stoplight and it's automatic, et cetera.

What, what are you seeing out there?

and a risk appetite that we can take on fully unleashing AI so that we can drive things at scale and sustain value?

Yeah, great question.

So I spend most of my time with chief data officers and, you know, maybe 10 to 15 a week.

And I learned so much from those interactions.

And oftentimes, you know, after, you know, every couple of months, I get to sort of, you know, lift my head up and sort of, Can assess what are the patterns, right? What am I seeing that sort of across that? And I think we're at a very unique and interesting moment in time.

And I think when I see sort of chief data officers, it's sort of in our continuum and largely, you know, they are sort of focused on.

one or three core problems and also describe those core problems.

One of the most important

So what is it? And then I guess what did companies used to do in the past with data observability?

Yeah.

Great question.

I'll start by talking a little bit about the history of software and product development.

Software and product development has been around for, I call it like the last 30 or so years.

And there's been a whole methodology and process around how to build reliable software products, reliable, secure, scalable.

And a lot goes into that.

I'm curious to hear if I take like a data infrastructure, and I were to like evaluate the health on that infrastructure, like, what are some metrics that you would suggest like an organization used to track that health evaluation?