Why Agents Break Traditional Databases | Andy Pavlo
MTS
November 9, 2026
Carnegie Mellon professor and ClickHouse VP of Database Research Andy Pavlo breaks down why AI agents demand an entirely new database layer.
SPEAKER_1
Hello, everyone, and welcome back to MTS. Today, we are joined by Andy Pavlo, VP of Database Research at ClickHouse, and a professor at Carnegie Mellon. ClickHouse just announced its biggest expansion yet, moving beyond a database into a full platform for the AI era. So Andy, welcome to MTS.
Andy Pavlo
Hey, thanks for having me. It's always good to talk about databases.
SPEAKER_1
Always great to talk about database. I know you're talking about, you are actually right now in Philadelphia, you are a professor.
Andy Pavlo
Pittsburgh, Pittsburgh, Pennsylvania.
SPEAKER_1
Pittsburgh. Other side, other side. Other side, not Philadelphia, Pittsburgh, Carnegie Mellon, of course. Of course, I get my cities a little confused sometimes. I would love to learn, before we jump into ClickHouse, I know you have so many announcements, lots of big things and I'm so excited to share all the big news. But before we jump into that, I wanted to start off a little bit with you and talking about your history with databases and how you ended up here.
Andy Pavlo
Yeah. So I guess I started as a professor in the computer science department, or assistant professor of computer science department at Carnegie Mellon University in 2013 After finishing up my PhD, working with Stan Szenadik and Mike Stonebreaker.
Then my research area has been primarily focused on databases. Databases is the most interesting topic you can have in, in my opinion, all of computer science because it encompasses anything you could ever care about in computing. Whether you care about algorithms, systems performance, programming languages, networking, operating systems, whatever you want to do, you can do in the context of databases, and people pay attention, and you can have real-world impact. So I joined ClickHouse this summer in August, and where I was asked to come along and build out their new research lab, just focus primarily on long-term exploratory research projects that eventually will hopefully make it into the product on the engineering side, but we can explore topics that maybe the product team can't do right away. Yeah.
SPEAKER_1
Man, that's exciting. So you're only three months in, and I want to get into that in a little bit. But for people who are just hearing about ClickHouse today, how would you describe it? Yeah.
Andy Pavlo
So there's two things. ClickHouse, the database system, and then there's ClickHouse, the company. So ClickHouse, the database system is a analytical database system that is designed to ingest huge amounts of data very quickly, and then be able to answer questions about it within milliseconds.
It's been developed since 2009 It was open-sourced in 2016, and then the ClickHouse, the company, has been working on the commercial Cloud version of this, and continue with the open-source version since 2021 So the ClickHouse, the database management system features a lot of the standard things you would have. Instead of the features you want to have in a analytical system to answer hard questions on large amounts of data.
The ClickHouse, the company, obviously, weekend is the primarily, the engineer team behind the open-source product. And then they also have the cloud offering that provides us some additional features and a managed service. And then the ClickHouse, the company as well, has expanded off into different other areas in the world, data processing and databases. So in particular, today's announcement is the general availability of their managed Postgres offering. They have a bunch of other products or capabilities now in the observability, security and AI space. And some of those things are built on the ClickHouse database system. The Postgres stuff can interact and integrate with the ClickHouse database system as well. Yeah.
SPEAKER_1
You brought up an important point. Database management systems are the backbone of how everything in software works. It's all tied to databases and having you and your background as a professor, you study this for a living, you teach about this for a living as well, is really interesting. I want to get into the last three months because you started at ClickHouse Labs this August. I want to know what happened in the last three months. What was the big innovations that came out of the lab?
Andy Pavlo
It's early. Research takes a long time.
I can't talk about too much about what we've done just yet because things are still cooking. But one of my things when I first showed up was go in this observation mode and look at all the different projects the company had going on the inside because they've done some amazing things. Some of those things I knew about are the reason I joined. I would teach my students these things because they put out a blog article and they had one research paper two years ago. I talked about some of these things. But then I didn't realize how many other amazing things they've been working on. They just sat down and wrote about it or done the experiments or looked deeper into what they're actually doing and trying to evangelize it. My early days has just been talking to so many different teams about, oh, we did this thing. Look at this. This is kind of cool and it cares how it compares against other database systems that are out there. We're really in the initial weeks, we're just soaking it all in and seeing everything that was out there that I could play with.
Read the other 17 minutes of this episode here.
$15 gets you this episode and 99 more. Credits never expire. After checkout, one link brings you back here to read it.
Not what you expected? Full refund within 14 days.
Thousands of transcripts fetched by people building searchable podcast archives
Fetch transcripts with code
The same credits work from a script or an AI agent. The demo key returns a sample episode in full, no card needed:
curl -H "x-api-key: pt_demo" \
https://spoken.md/transcripts/1000651996090Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.
From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.
Using your own key:
curl -H "x-api-key: YOUR_KEY" \
https://spoken.md/transcripts/1000800867334