Demo Is Not Production, with Ajay Nair, GM of Elasticsearch and Platform at Elastic

Ajay Nair (0:00)

Every team can get a demo working, but demo is not production. Blast radius of them doing a wrong action is so much worse than you giving me a wrong answer. Look, as the search guy, I will tell you AI has a search problem. Don't burn your tokens if there's a simpler way to do it.

Right?

Joel (0:23)

Just curious, where are you calling in from today?

Ajay Nair (0:26)

Seattle.

Joel (0:28)

That's like the heartbeat of technology right now.

Ajay Nair (0:30)

It feels that way, other than every once in a while, football takes over and then it feels like most of the time tech's what drives the heartbeat.

Joel (0:41)

How long have you been with the company?

Ajay Nair (0:43)

I have been here two and a half years. So my role here is driving the Elasticsearch AI and Platform portfolio at Elastic. We started out as an open source search engine, and we work on observability, search, security.

I've been here now two and a half years. I spent a decade at AWS before that in various cloud and serverless platforms, and then about eight years at Microsoft in big data and actually search as well, although it was the other form of search before that. So this whole waves of search cloud technology and to some degree, AI and ML has been very much the place that I spend a lot of my time in. Elastic is the best place to bring all of that together.

Joel (1:27)

That is amazing. The first time I needed complex search, I was looking, talking to people, and then I found Elastic in the Heroku Marketplace.

And I started learning it from that. And before that, I would just roll my own and use other libraries. I needed this specific functionality and I needed it to be fast. And I found their tool. And so in my mind, that's how I know you. And then when I was preparing for the interview, I found out that you guys have grown. You're in observability, you're in security. You're like way more than just the search.

Ajay Nair (2:00)

Yeah, no, I might.

The running joke inside is that Elastic's gotta glow up. You know, a lot of people associate Elastic with that core sort of search functionality. And it's absolutely the big foundation of everything we do. But you realize, as you do with today's world, that search plays a big part in many things, right? And there's a many time when you want to find the right bit of information to make the right thing work, and observability, security, and now increasingly a bunch of AI stuff. Like Elasticsearch has found a lot more users since then, for sure. Last 15 years have been a blast for the company. How was your experience the first time you put it on? Are you still using us? Still playing around with it?

Joel (2:39)

Oh, I don't even have the skill to build that because I just would send it to your API. So I can't talk deeply technically about how these things work.

But I had a problem. I had a request for a feature. I found that as the solution. I said, oh, okay. Now I just use that from now. It's just like how you found anything as an engineer. When you found jQuery, you were like, oh, this is great. And you just use it, move on.

Ajay Nair (3:04)

Yeah. I suspect the right tool for the right job is going to be something we'd talk quite a bit about today for sure.

Joel (3:10)

Yeah. And I'm curious too because you get to, I mean, 50 percent of the Fortune 500 use you guys. You're huge, you're in the infrastructure play for basically every large company. But you're at the top, so that means you're having conversations with other leaders at companies about AI, the production gap.

I saw the stat, it said 41 percent of new tech spending is going to AI, but under seven percent of enterprises have it in production. What's happening there?

Ajay Nair (3:41)

I'll actually probably start with an honest conversation I had with a customer just recently. They were evaluating one of our capabilities called Agent Builder that allows them to build production agents using elastic data. And they went through the POC, everything was fine, and they walked away saying, hey, the ROI bar is not clearing for us because the cost benefit of it doesn't quite work. And the underlying thing that was really illustrative for me about the whole thing was, every team can get a demo working, but demo is not production. I think that's the biggest failure mode I'm seeing with people who are thinking about AI.