Agents in the database (Interview)

27:36And I think that's been fascinating for me because, I mean, I think it's totally changed. I mean, the core job that database does has not changed. But I think the way you build the business has totally changed from being more sales-led to being more product-led. I think that's been interesting. What was the original challenge you faced to even consider creating Timescale? What was that moment? When was that moment? We started off as a company building an IoT platform, Internet of Things. And it's like 2015. Yeah. I mean, back then, I just spent 10 years in mobile. And I remember thinking, wow, mobile was really exciting.

28:23But around 2014, it started to get a little bit boring. Pre and post iPhone was exciting. But then 2014, you're like, okay, I have enough apps. I'm like, what's the next thing? And IoT felt like the next wave of computing. And so we started off building what we thought the market needed, which was a data platform for IoT devices. And that idea was moderately successful. We tracked over 100 ,000 devices. We raised a seat around, built a small team. But we needed a database to store all this data. And we were using a time series database. We were using a relational database. And I remember at one point, we wanted to, I think, sort the console by uptime.

29:02but then show all the device metadata. And what should have been a simple SQL join ended up being like a two-week engineering sprint because you have to connect these two siloed systems. And I remember thinking, oh, this is awful. This sucks. And one of our engineers said, hey, I could build this on Postgres, but it'll take me a month. And I was like, okay, cool. You're an optimistic engineer, so it'll probably take you three months. And 12 months later, we had this database. And I'm trying to sell this IoT platform. And look, as an entrepreneur, you learn to listen to signals when someone is really engaged.

29:43And when someone's like, oh, yeah, that's cool. That's not cool. But when someone's like, wait, wait, hold on. Can you tell me that again? You're like, oh, okay, there's something there that's interesting to you. And so here I am trying to sell this IoT platform, meeting with the German head architect of this large shipping logistics company. I'm meeting them in Mountain View. And I'm telling him about this IoT platform. And he's like, look, there's so many IoT platforms out there. What makes yours different? And I was like, oh, we built our own database. It does SQL skills for time series. And then he was like, wait, wait, hold on.

30:16Wait, can you tell me more about that database? And I think that's when I realized, wait, this is solving. this is actually solving the right problem. And that's how it became a database company. So yeah, long story short, we kind of scratched our own itch and realized other people had the same itch. But again, we are students. We listen really well to the market. We try not to be dogmatic. And when someone says that thing's not interesting, but this thing is, we're like, hey, we're not here just to build that thing. We're here to solve a problem. If this solves a bigger problem, let's explore that.

31:01Well, friends, I'm here with my good friend, Chris Kelly over at Augment Code. Chris, I'm a fan. I use Augie on the daily. It's one of my daily drivers. Now, I use Cloud Code. I use Augment Augie. And I also use Amp Code and others. But Augie, I keep going back to it. And here's where I'm at. I feel like not enough of our audience knows about Augment Code. not enough about Augie the CLI it's amazing I love it what can you share yeah we often say Augment is the best coding assistant you've never heard of and that's both frustrating as someone that works there and it's like very proud of the work we've done but also like inspiring like we want to go and and sort of punch above our weight because like we aren't anthropic and we aren't open AI and so the quality of the product itself you know with our context engine once you do touch it people are like just blown away by that and so like that keeps me going every day so not to better lead here, but this is a paid spot.

31:56You are sponsoring this show to get this awareness. Now, at the same time, we're selective and I love to use your tool, but there is in the world. So a lot of developers look at the space and they say, okay, well, how long can this work? How long is this sustainable in the case of cursor or windsurf, or you pick the name and you think discounted tokens help me shape a lens for our audience. I think it's a lot of awareness, right? Like Like Cursor got a lot of publicity early on for like fast revenue growth, which well deserved. I think, you know, frankly, some of the media gets the story wrong.

32:35And that like, if I gave you$1.50 for every dollar you sent me, I'd be the fastest growing startup in the valley. And so when you're selling discounted tokens, yes, of course, you're going to grow very fast. But all that money plus more goes to the model providers. So I think the real story is the story of Anthropic and, you know, being an API provider. I think the market has just moved so fast and there's so many pieces of competition out there that it's just hard to get noticed. So, friends, I love Augment Code and I love using Augie and I highly recommend you use it. I love using Augie. I can hand Augie a well-defined specification, a well-defined PEP, as I call them in my world, an agent flow, and it executes flawlessly.

33:17So the cool thing about Augment that I love most really is that context engine. And I can hand it a task and it can just churn away on my well-defined plan and just never bother me and accomplish the mission. It is so cool leveraging the latest models, the context engine, and all the fun things behind the scenes in that awesome CLI. So yes, go try it out, augmentcode.com. Right in the top there is a CLI icon, a terminal icon. Click that, install it, and change your world. It's going to be awesome. AugmanCode.com

33:56Today, you're not called Timescale, though. That's right. And so there's a name change of recent, which I think is challenging, but it kind of maybe shows the evolution. What's the evolution of that discovery, that 12-month discovery, scratch your own itch, IoT company to Timescale to now be Tiger Data? We started off thinking we were building a time series database for IoT. That's where we started. And so when we kind of pivoted to becoming a database company early 2017, we called ourselves Timescale. That seemed like a good name for a time series database. I like it. That's cool, right? It's pretty self-explanatory.

34:33And we actually saw there was a big demand in time series that was more than IoT. some in finance and events, and then soon crypto and some other areas. But over time, again, we're going through this journey with the industry. We started off as an enterprise sales motion. At the beginning of the pandemic, we went all in on cloud, stopped selling on-prem, and said, hey, we're going to be a cloud company. And yeah, at that point, we were running databases for customers, and we got to see a lot of data. And I saw, wow, there are a lot of companies who are like 20 people spending 50, 100, 150K with us.

35:12And I remember asking like, hey, what are you doing? Like, you know, what do you use this for? And I remember one of them was like, hey, I think of you as a better Postgres. I don't think of you as just a time series database. I think of you as my main, you are our main database. You are 50 % of our cloud spend. You are our main database because you're a better Postgres. And then we realized what we had built was not just a better time series database, but a better Postgres. And we kind of expanded into a better Postgres theme by adding vector support and better native AI support. And we got to the point where this year, we realized, hey, we keep calling ourselves Timescale.

35:50And people view us as a time series database. But we had already become something more than that. It's as if Amazon.com has started off as Books.com. And you're like, hey, Books.com, we also sell CDs. People would be like, yeah, but you're at books.com. Hey, we also sell socks. Yeah, but you're books.com, right? And that's what timescale was. Timescale was like, yeah, you do AI, but AI with time series, right? And we're like, well, no, AI. And I think our first attempt was to try to make timescale mean more than time series. Like when it's time to scale. I don't know. and I'm not sure if you've ever seen the movie Mean Girls but there's a line in there that Mike has quoted where he goes it's like trying to make fetch happen like it's not going to happen like we're trying to make this term stick which wasn't sticking and so we're like hey you know what we need a new name because we are Amazon we're not books.com and but we're not pivoting we're actually changing the name to properly reflect who we'd already become Yeah.

37:01And tiger data was like the perfect name for us because we, our mascots always been a tiger internally. We talk about tiger time. That's our all hands. We talk about state of the tiger monthly. It's my monthly presentation to the company. New people who joined the company are tiger cubs and yeah, like we've internal tiger mascot. And so the name tiger just worked because it was who we already identified with and it, you know, to our existing customers, it looked like, yeah, like it's your same logo. Cool. I get it. And some new people were a little confused, but we were like, you know what, like in 12 months, you will not be confused.

37:41And this is a right thing. So that's why he made that change. Was it scary to, to, to get to the point where you're like, you know what, the only way forward is a name change, nothing else, just the name. it's you know it got to the point which probably means we'd waited too long it got to the point where it was painfully obvious to us yeah and and i and i remember and i told the team look this is not going to solve any problems but it is going to remove an anchor that is holding us back like it's still on us to kind of move the ship you know yeah but like it removes an anchor i think the key thing is you have to stick to the decision and be like, yeah, it's going to be messy.

38:24It's going to be maybe a year of people being like, who are you again? And we're like, yeah, we're Tiger Data. I actually think the name is catching on faster than I expected. I think the name changed when it's smooth as we could have hoped. And I think people, TimeskillDB, I think still has more brand awareness, but I think Tiger Data is catching up. Yeah. Some people made fun of us, but like, whatever. Well, haters hate, right? Haters hate dude. Yeah. Haters do hate. So, you know, our audience knows this. You've been a sponsor for a bit too. And, uh, you know, I mentioned maybe in the pre-call, maybe in the early part of it, uh, Isabel is someone who works for you.

39:07Uh, and we've known each other for years since the MongoDB days and we would have conversations and this is when you were timescale. And this is what I do whenever I sit down with a brand And I think about how can we help them reach our audience in a way that is informative, educational, and just something that helps them be curious and try it if it's something that fits in their world, essentially. And, you know, I always do this version of an investigation to some degree. Like, look at your homepage, look at your products. What are you doing? And, like, things just didn't pair up, especially, like, when you mentioned with AI.

39:44Like, yeah, it's time series, but it's also just Postgres. and it was hard for me because i was i would tell isabel like hey i would write her you got to change that that headline on your home page it just doesn't just doesn't work like i did yeah you're something that's not fitting here and so when i saw the name change you know reflecting on your your uh your mention of it seemed natural essentially when i saw the new name the first thing I did was email Isabel and I was like, listen, that's an awesome name. We got to talk. When are you, when can you make time kind of thing? Cause it'd been a few months since I talked to her and that's what happens when you do a rebrand and change your name.

40:24You, you rethink your model, rethink your brand and you come back out with a new plan. And that's what I saw. And so the moment, just to kind of reiterate the moment I saw this new brand tiger data, I was like, that makes total sense. The tiger's there. The logo's awesome. It makes sense. I'm glad you felt that way because I feel like people who knew the company mostly felt that way, which was like, we didn't even change our logo. Our logo stayed the same. We just changed the name. And people were like, yeah, no, they get it. And by the way, tigers are cool. Tigers are cool. Tigers are cool. Right?

41:00They're cool. They're fun. What I liked too, when you did this, was the logo animated from timescale to Tiger Data. Logos stayed the same, right? The mark that you have. I thought it was a nice little subtle touch. That was our marketing team. Yeah. I also, you know, I'm not a, I think a lot of people like dark mode. And I think your previous site was all dark mode. And you may even offer a dark mode version of your site. I don't know, because I see the stark white version of it. it doesn't bother me. Like the yellow and the white and the black, I just, it's, whoever was in charge of that process did a great job thinking through the core of who you are and how do you come back to the side, not being completely different, but being different at the same time.

41:50Yeah. No, I, that was, I mean, it was a team effort, but marketing team drove, drove that kudos to them. But, but I think it definitely gives you like some thought on like, Hey, does the name really matter? And I think where I'm landing right now is that the name may not help you, but it can hurt you. I mean, there are two schools of thought. One school of thought is you name yourself books.com. Other school of thought is you name yourself Amazon, which could sell books, right? And I like the clarity of timescale because it was like, hey, time series. But in hindsight, it was limiting. While like Tiger, dude, Tiger could be anything.

42:30Tigers are cool. Tigers are cool. So one of the things that you have now, I think, is this burgeoning idea of agents in our databases. Yes. Right. And, you know, I think I'm at this point where there was this, this, this arc of acceptance. I would describe it of, Hey, come on now. AI hype. I'm over it too much, this and that. And I think the game changed when cloud code changed the game. That's where I really think it happened. Like I had been a user of chat GPT, like many people, I'd been the API to do something in chat GPT and copy it out somewhere else, whether it was written material, whether it was an idea, whether it was a framework in terms of a thought framework, or maybe even a bash script because AI is pretty good at bash scripting.

43:24And then you kind of get to this other side where you're like, you know what? Wow. Agentic is, is really revolutionary. And I think cloud could really change that game from the new browser or the new destination point for developers, it is still in the IDE, right? And that's still taking place today. But I think you're seeing this shift to the CLI that is just truly revolutionary that now the terminal, it was never not cool, but it's cool again, right? A lot of folks are hanging out there. Cloud code kind of put that on the map in a way. And then a lot of folks decided to follow that direction.

44:04And I imagine that's kind of what you thought too, is like, I'm, I'm probably doing, you're probably doing a lot of agent decoding. You're probably playing with side projects. You're getting curious again. And you're like, well, the next best thing is to, how can I just talk to my database? How can I just put an agent in my Postgres? Talk about that. Yeah, I think, yes. I had a very similar experience with cloud code. I remember one of my friends talking about agentic workloads in 2023. And I remember thinking, like, what are you talking about? What is agentic? What does that mean? And now I use the word agentic at least five times a day.

44:43Yeah. And Cloud Code was that moment. Cloud Code. I mean, ChatGPT was cool, but it was like a party trick. You know, it was kind of cool. I could edit for me. For developers, it was a party trick. I mean, for everyone else, it's pretty dang powerful. But as a developer and ChatGPT, it's not the right interface. Cloud Code. Yes, but not in a web browser. I remember building, because I was trying to talk to a friend. I was showing it to a friend. He was like, oh, can we build an app that tracks pushups? And I'm like, I don't know. Let's try it. And I think 45 minutes later, we chat, we had gone out and found the right computer vision library and other stuff.

45:24And I had a mobile web app that would use computer vision to detect if you're doing a pushup or not. And it didn't get it always right, but it got it right at maybe 80 % of the time, which was pretty good for 45 minutes of work. And I remember thinking, dude, I can build anything. I can build anything now. And I'm with you. I think Cloud Code, it felt like, wow, this is actually an agent doing work for me, writing code, making decisions. I can steer it, making me more productive. I remember going to a social event that night and I couldn't even talk to people because I was so excited. I went home early.

46:06I was like, I need to get back to Cloud Code. I'm not feeling well, y 'all. I got to go. I swear to God. Well, I'm trying to talk to people about Cloud Code, and they're like, are you okay? I'm like, no, this thing is amazing. You have to try it. What kind of party was it? Was it nerds or was it normal people? It was part nerds, part normal people. Okay. But even the nerds were like, I mean, Cloud Code had just come out. I think it was, they were still like, yeah, I use Cloud. I'm like, no, not Cloud. Cloud Code. You know? Yeah. You know? No, I mean, you know what I told you about that experience in 1996 using the internet for the first time?

46:44Like, that's what this felt like. Like, number one, it felt like, I don't know what this is, but I want to be a part of this. Like, this is fun, right? But also, it brought me, and I know a lot of my peers have feel this way. it kind of brought out this like child childish sense of wonder that I think like years of being of an entrepreneur had kind of beaten down and it kind of brought it back to the surface to be like this is fun this is fun the way technology should be fun where you're just like I just want to tinker with this I don't know what I can build I just want to build things explore this new world together and yeah so I'm really excited I mean I think as a company, we're excited because we see this trend that 80 % of Claude code is written by Claude.

47:36The majority of new software over the next 12 months will be written by AI. We already have customers who tell us that 70 % of their code is written by agents. And you just look at this and you say, okay, if that's happening, then what else is happening? Well, that means the surface area of software development needs to evolve for agents, right? The surface area of databases. So now Now databases are serving a new user. They're not serving a human. They're serving a human using an agent. So then you ask yourself, okay, then how does a database need to evolve for that? Well, number one, it's less of a GUI.

48:09So you remember like databases went from on-prem to SaaS, like cloud. But now you're going from SaaS to, I don't know, MCP, CLI. It's like the interface is totally different. It's no longer like clicking, right? It's more like commands. And there's some other things too. Like you want databases to boot instantly. You want to be able to fork and create sandboxes quickly in a safe way, a cost-effective way. You probably want native search, native memory. Yeah. I mean, so I don't know. I'm just excited as like the little kid in me who just loves building things. I'm excited. But also as a company, it's like, yeah, this feels like a problem someone should solve.

48:44And the answer is probably Postgres, and we might as well solve it. This lyric lands for me in this moment. It's from, I had to look it up. I don't know this by memory, but it's from Marvin Gaye and Tammy Terrell. Ain't nothing like the real thing, baby. So when you're in ChatGPT and you're just ideating the future of what your software thing could be, if that's where you're kind of hanging out at, versus that experience of in Cloud Code doing the same thing, but it's making the thing. It's making the real thing. And you're seeing the real thing change. And maybe you're doing Git commits. Maybe you're doing spec driven or document driven development and you're thinking a ton and you're writing it down or it's writing it for you or it's writing the markdown.

49:33But the ideas and the vision and the direction is coming from your lived experience and your wisdom and places you've been and valleys and cul-de-sacs you've been down. You're like, not going there. And it may try to take you there, but you're like, no, no, no. Hang on. Let's curb that back. Let's go here. but you're seeing the real thing be made that to me was the was I think the moment where I was like that's pretty wild the real thing and you could be like okay cool commit those changes and push them upstream and so you're like oh it's actually fitting because when I use Replit or Lovable they're fine but I was like okay cool but I don't build this way maybe other people do I don't but now I'm like yeah I build this way like yeah commit the changes pushed upstream like yeah cool great show me the diff all right you know like the speed you can move at i think is is is kind of wild you know i laugh about this and i don't know how often you laugh about this but i love when it makes a plan it's like well this is week one this is week two and this is week three meanwhile four hours later the thing's done you know or an hour later whatever the number is that is always funny yeah i like how it manifests its time ranges but i'm like it's kind of funny you know it's still making a plan or it's making a plan with you and the range of its timeframe is, is maybe it's actually accurate to how it should be if it was done with a team of humans versus a team of one with an agent and an idea.

51:05And I think that's kind of wild. It's just the, or maybe it's just under promising and over delivering. Cause that's what it's like. Maybe psychology, something's going on there. Even the, even your mention of, um, you know, we had a conversation, you and I a few weeks back. And you mentioned the same thing. You mentioned this childlike play and that's kind of the definition of flow. Like flow state is that when you're in that state of flow, that state of play, a lot can happen there. You stay more engaged. It's kind of addicting in a way. I'm a big fan of that idea, the concept of play. When something feels like play, that's when you know you're like you're in it.

51:44Like you're in the flow. You are doing the thing you should be doing. Yeah. And if I ever write a book, I think the title will probably be play because I feel like that. I think that should be the driver. It's like, look for things that feel like play because then those are things that really like kind of fit who you are. And if you find something that feels like play and also lets you kind of support yourself or your family, yeah, dude, that's the jackpot, right? Yeah. But the AI story kind of reminded me of something. When the personal computer came out in the late 70s, people struggled how to describe it, probably similar to how we struggled describing Chad, GBT, and Claude today.

52:30And Steve Jobs, who was probably one of the best tech marketers of all time, either he came up with this or Apple came up with this, but they would talk about how a computer is like a bicycle for your mind. which is like a kind of a weird like today that feels like a weird analogy but like i i get it right because back then you're like hey this thing this is like pre-spreadsheet right hey this thing like we don't know what the key use cases are but kind of like how a bicycle lets you get from point a to point b faster like this is just making you better and that's how i feel about cloud code and there's all this discussion about like hey our agent's going to replace us i mean maybe some of the things we do but for most of the things i think it'll be like a bicycle it'll just help us get from point a to point b faster yeah i don't know i think that's part of the joy is being like hey this thing that used to take me a week or i couldn't even do before now i can get done in an afternoon totally throw it away and start again yeah like the next day Try to juxtapose the timeframe of IoT company using Postgres, needing time series, engineer, developer, employee, whatever comes to you and says, I could do this in Postgres.

53:47I could do it in a month. You're like, yeah, you're ambitious. You can do it in three months, 12 months later. That whole story. Juxtapose that time to idea, to CLI. And we haven't even talked about the CLI yet. but getting to this point of agentic Postgres and building what it is. How fast did you go from idea to using the thing? I mean, a lot faster. A week? A day? I don't even know, man. I mean, I think this is the theme in the AI land is just ship. Just ship? Just ship, right? We're a database company, so we have a slightly higher bar than maybe a classic app, but like Edge of Ship. Yeah. I mean, there's definitely a shift between it used to be hard to build, but now it's a much easier to build, which probably means that the more of the bottleneck then becomes a distribution and awareness, which is like, I think something similar to what, you know, the iPhone is.

54:52It's also in the bottleneck though, distribution awareness. It's just exacerbated now. Yeah, but I remember once upon a time, pre-AWS, you probably remember this too, which is like, to start a company, you needed a server rack, maybe a couple of Sun Micro system machines, Oracle 8i database. And just to get started was such a high bar that a lot of people couldn't do it. I was actually going to say, I think it's similar to what the smartphone did to photography, right? Once upon a time, being a photographer was really hard, even if you look at before digital cameras, right? And you'd only take so many shots.

55:34You have to develop it. It would take time. It was hard to get right. You would have to actually apply physical filters and physical lenses to change the thing. And with a smartphone, everyone became a photographer. And what that meant was that like in the past, having great photos would set you apart. But now it's like anyone can create a great photo. But how many followers do you have on Instagram? So kind of again shifted the burden from photography skill to like distribution. And I feel like we're going to go through something similar with software. How much do you know about the, I suppose, the details of the CLI and the MCP server that is in your product?

56:18Because I felt it was pretty, when I look behind the scenes of the open source of that, you know, it's elegant the way you've married a CLI and an MCP server in a single binary. That's right. That's pretty wild. I think the design is, I had to borrow it, honestly. That's great. I got a couple of things I'm working on that are one is a DNS resolver. And I want to like talk to the resolver. And so the obvious answer is to have an API, right? But then you also have to have a CLI that talks to the API. But then I'm like, well, I'm actually using an agent. So I'm taking a playbook from Ajay and Tiger Data and Agentic Postgres.

57:03and, well, the same CLI can offer an MCP server running at the same time and you can talk to it and get back JSON for the agent or get back CLI response for the human. Like, the design there was so elegant. I want to know what you know about that design. I know the team that built it is really smart. You know, I know. I think they looked at a couple different options. I think the way it works right now is that the CLI will run a local MCP server, which then speaks to a remote one that we have. And the remote one is where we index, you know, our MCP will allow you to do things with Postgres and Tiger, but also lets you, you know, search the Postgres documentation, download best practices.

57:51And so we put that on its own kind of remote MCP server. I think MCP in general, it's such an interesting area. And I feel like a lot of people hate on it because it seems so simplistic. But I think just the idea that people talk about in the physical world, the idea of a design affordance. You know this concept? It's this idea that when you see something, when you see this mug and you see this thing, you know you can hold this. Right. Right? You know, it's kind of like, you know, a door handle. You're like, I know I can turn this. There's one reason why Tesla door handles trip some people off because they don't see the handle.

58:29What's going on here? Yeah. Actually, I think that's bad design. But in affordance is when the tool teaches you how to use it because it's designed the right way. Like games, like Mario. The Super Mario Brothers, the first game was like that was why it was. That's right. Level one was level one. It was teaching how to play level two and three and four. And so this is where MCP, to me, is not just an API. It's essentially a way to expose tools for agents. To essentially think about like, hey, we used to make these handles for human hands, but now we're going to make them for agent hands. And that is a really growing art.

59:13Anthropic who developed MCP, they're still learning best practices. I think that's really interesting. It's to figure out what proper tool design looks like, how many tools is too many. I think right now tools are mainly a wrapper around APIs, but I think it'd be cool if they were actually doing the job, not just the API. Yeah. So I think the team's done some good stuff. There's really other some good stuff that we're going to release soon. But I think MCP design, I think this idea of designing for a new user, the agent is, I think, a really fascinating area.

1:00:04so our friends at framer are fans of this podcast and their sponsor you know most design tools they lock you behind a paywall well framer flips that script it is a free full feature design tool that does something that most site builders cannot. It's actually designed for designers and Framer already built the fastest way to publish beautiful production ready websites. But with design pages, they've redefined what it means to design for the web. This is not a Webflow clone or a WordPress competitor. It is a true design platform, vectors, 3D transforms, gradients, wireframes, all the tools you actually use and they're all free unlimited projects unlimited pages unlimited collaborators and here is the kicker you design you iterate and you publish all in one place there's no figma handoff there's no messy html imports there's no tool switching and for designers and developers who are tired of the tool switching this whole dance you got to do to create social media assets to create campaign visuals icons entire sites you can do all this now without leaving framer this is where ideas go to live start to finish so if you're ready to design and publish in one tool start creating for free today at framer.com design and use our promo code changelog for a free month of framer pro again framer.com design use the code changelog for a free month of freemapro rules and restrictions may apply well we've heard the term dev tools and now we have agent tools that's right right that's how i think about things like if i'm done and i've done a couple toy clis just to play with them one for prox mox i'm a home labber i don't know what kind of nerd you are but i love to home lab and one of the epicenters of my home lab is some version of a Raspberry Pi somewhere or an Intel NUC or maybe a slightly bigger one, which is a NAS, which is usually true NAS.

1:02:14And the next best big thing in the network, for me at least, is a Proxmox machine, which is dedicated, where I can do different things, whether it's a service or tinker with something or play with a new version of Linux or just DistroHop or DistroPlay, however you want to describe it. but as I'm, as I've explored this world a bit and I've understood what it's like is I'd never really built many CLIs before. And, and I was like, well, the, the next thing you want to do there is really just, you want to enable now that we have agent decoding and it's, it's almost ubiquitous. So when you build a tool or you build a new CLR, you think about a new way to create an API with a CLI that lets you talk to a service.

1:03:03it's, it's, you can assume or presume pretty, pretty good with pretty decent accuracy that the person may be using an agent, likely using an agent, or at least has one nearby and is, you know, working towards being okay with that. Some people are still not okay with that. So I was like, you know, the, it only makes sense now that when you did, when you design these tools, you can't just designed the old way, which was, well, I'm designing this obviously to solve a human problem. So it's going to be a human using it. Now we have humans that are still there. There's still the human in the loop or on the loop, depending upon the framework.

1:03:46But that human is leveraging one or many agents or sub-agents to automate and to do different things in their tasks, whether it's designing the plan, whether it's writing all the code and they're reading the code, they're still doing the code review. Maybe they're doing 20 % of the coding because the agent does 80 % of it or 90 % of it or whatever the number is, you know, likely this person has an agent. And so we now have to look at the way we build tooling in a way that's like, okay, I used to do it this way, human only, but now the human has a friend, an agent, I can design for both of them because that human is expecting their agent to have agency on their behalf as they command it to do the task.

1:04:30And the agent speaks a different language, reads a different way and interacts with a command line or an API in a different way. And that's where MCP or even like cloud plugins come into play. How do you feel about that world? How do you feel about like agents in our Postgres? And now we can actually talk to our database. I think, I mean, agents will be in everything because I think it is someone helping you use the thing that you're using. And I mean, you may not have an agent like inside the database itself, but you probably have an agent inside the system that's running the database. That's one thing we're exploring right now on Tiger Cloud is having agents that help you.

1:05:15No, but it is totally right. Like I think, again, it's not replacing the human. It is the human doing the thing they need to do, but now they're doing it through an agent. And so now you're saying, okay, like the job to be done is still the same, but the thing doing the job is different. And so it's going to need different things than what the human needed. Absolutely. Yeah. I mean, I think like some places we're looking at is like, you know, what are some of the menial tasks in a database that you don't really want to do? Right. Okay, cool. Like maybe you start there. Right. But also, what are some tasks that you can't do?

1:05:58You know, maybe an agent can because agents aren't limited by time and space. You know, they can easily paralyze and run things and shut them down. When I look at what you've done with, you know, agentic Postgres, which I think it's, I'm not sure if you're trying to brand that term. I think it is the first time I've heard it. I wonder if that would just become the thing. And it's a product named like Xerox or Kleenex. And I won't be using Tiger Data's agentic Postgres. I'll be using somebody else's because there are more Postgres database providers out there. yeah i wonder if that's the case but when i look under the covers of this thing and what you've done it's the first time i can look at something and say now i can actually talk to the database now i can actually sort of command it to go do something and it goes and does it it could even be a branch or a fork or whatever terminology you're using to say give me a safe area to play in that's just never this has never been a thing before you know and now now it kind of is yeah Yeah, I mean, the CLI, I mean, I think CLIs have been useful for, they've been a thing for a long time, right?

1:07:07What, 15 years, maybe? Yeah. I mean, I don't know. If you look at Unix, maybe it's forever, you know? They're on the way back. I feel like CLIs are like the next frontier in a way, really. I feel like the next year or so, we're going to have this massive swath. And maybe some will live and some will die, but that's going to be the delivery mechanism. So, yeah. I think the key thing is the CLI is a clear program that agents can use. So, okay, so when we built up Tiger CLI, like it's written in Go, it's distributed as a single binary. It essentially provides a unified control plane for your database on Tiger Cloud.

1:07:41And so, you know, it covers, you know, auth obviously creating, starting, stopping, forking databases, connecting databases. But like what's really, I think, nice about it, kind of like you pointed out, is that it has MCP baked in. So, you know, you can run essentially a local MCP server through the CLI that your cloud code or whatever can talk to. And now your agent has access to all these basic commands. It has access to starting, stopping, forking databases, connecting to databases. We've also it also proxies another MCP server that we host where we have indexed the Postgres and Timescale and tiger documentation.

1:08:31And now we provide semantic search and tech search on there. So again, it's not just a CLI. It's not just a CLI-FCP server. It's a CLI-FCP server that actually makes your agent smarter. I don't know. Like I would say, I'm not sure if this is true, but it probably is. This is probably the first AI native CLI for Postgres. Like it's not just like, I think other Postgres CLIs are just like API wrappers, But this is really thinking about bringing the CLI commands into the agent, but then also giving agents the Postgres context it needs, best practices, documentation, so that it's instantly like an expert developer.

1:09:14Yeah. I mean, I know a lot of developers hang out on the command line, but I feel like to some services, the CLI was not so much an afterthought. It was sort of you have to do it, but not a lot of people would use it as the kind of primary interface. And now that we have agents in the command line, I feel like it's table stakes. If your service doesn't have a fleshed out thought through well-defined API so that you can also have a thought out well-defined CLI. And then, you know, maybe MCP lives and maybe it maybe it dies because agents can just sort of map around. I know Cloud has plugins and things like that.

1:09:56So you can sort of use Markdown to teach it of all things, not TypeScript, not Go, not Rust, but Markdown. like words uh maybe mcp there's there's talk about you know is it a fad is it not i'm still not sure because how how sure can you be when things change on the weekly even on the daily but just this idea that now the new frontier for a service is a cloud is an api and is a cli at its at its best and maybe mcp lives or dies but for sure you got some sort of plug-in system that you You can teach with a slash command in the CLI that most people are using. If it's open code, if it's cloud code, if it's Augie, if it's AMP, like you pick your agent.

1:10:43That's right. You know, or codex even. They're always last on my list for some reason. But, you know, the moment you give them that slash command, now the agent has access to the CLI. If it's got great help docs, so dash dash help on all your commands. Now the new frontier is like this really well fleshed out CLI. I feel like that's where so much attention is going to be going, if not already there. No, it's a great point. And I think, and again, like, hey, is MCP going to be the end all protocol? I don't know. But there will be a protocol. And I think that's the point, right? That's the point you're making, the point I'm trying to make.

1:11:21And why do you need this protocol? Well, number one, you need a way for the agents to interact with your APIs. But number two, you probably want to give them access without giving them the full keys to the kingdom, right? You want to give your agent just enough power, but maybe sandbox in a certain way. Maybe they can do some things, but not others. And I think this is where MCP, your agentic APIs really play a role. And then the CLI is another way to do that. CLI is another way to give essentially a tool to the agent that is properly defined and what it can and can't do. Yeah, no, I think it's really interesting.

1:12:00I think the team's done some really cool. I think this docs thing, and I'm not sure if I'm beating a dead horse here, but I think what they did with the docs, I thought was especially clever. I haven't seen anyone else do. It's like, hey, we have an MCP server where we have indexed the Postgres documentation and the Tiger documentation. And we've indexed it for text search and semantic search so that your agent can effectively get a PhD in Postgres.

1:12:30We've done the test. You can create schemas without our MCP and then with our MCP and then have Claude compare the results. And with the MCP, it's just a much better schema. I think that's been really cool. I think it's almost a variant of the skills that you see in Claude. But I think that's another element of agents that I think is underexplored is the fact that with the right markdown, they can become experts in anything. Did you see the talk from Barry Zhang and I think his name is Mahesh Marag? They're both from Anthropic. It's the first time I've seen. Well, I guess the first time I thought I would see anybody talk on stage for an hour about markdown.

1:13:15markdown they introduced i think i saw the tweet about it it was like a packed room or something yeah like yeah like an hour-long talk about using markdown uh you just wouldn't expect that right that's not the language you that's not the language i thought would be front and center now i love reading markdown uh i personally prefer markdown i've been a obsidian user for years now and so I'm all about flat files, nothing that's sort of obfuscating my pros from my source, and Markdown is that. But it's the first time I thought I'd ever see somebody on stage presenting a revolutionary new feature that is now used by so many folks now.

1:13:59I'm starting to use plugins. It's Markdown. It is wild. In fact, I have a friend working on a Markdown editor for Vibe Coding. you know like not for note taking right but for like vibe coding no you know it's it's a new world man you know maybe we'll find that's the weird thing with these agents is that in some ways they are very similar to us you know you don't teach it using machine code you teach it using text in Markdown. Words. Words, man. Words. Yeah. It is wild. I do feel that. I also think when your discussion earlier, I had a thought about the terminal, about how more people are using the terminal, limiting the terminal.

1:14:53There's probably a business idea out there, if I'm sure someone's working on it, of building the modern terminal. I think, well my favorite and I'll give them a nod because they are my favorite is Warp. I think they've been, gosh I recall talking to them almost five years ago and it was the future of the terminal I want to say. The terminal of the future I believe was the title we shipped on that. And Zach Lloyd is I think he's a solo founder even of Warp and Warp is my daily driver so that's what I use when I'm in a terminal I'm using Warp. Now I know Mitchell Hashimoto and Ghosty is awesome I love it I'm so thankful for what they're doing especially what Mitchell did with the you know the foundation kind of piece to it to kind of give Ghosty a home forever yeah but something about something about Warp just gets me and so I just like Warp so I think they may be ahead of the puck in terms of that yeah well it sounds like Warp did Warp start before this whole AI thing oh yeah Yeah, they were.

1:15:59It's probably two years before, I want to say, at least two to three years before GPT-3 made people think that emails will forever be written by machines. And right now, I don't know. Right. Now I mean, like, and now, right. You know, can you summarize this into an email form for me? I will edit it and I will remove what I don't like or whatever. However, that's, that's, it's not how I write all my emails by any means, but a few that require attention or care or particulars. Why would I write all that out? You know, I can have, I can have the thought and the thought process with the agent and then it formulates all the thoughts I've just given it into a cohesive form.

1:16:42And I can move along to my podcast with a J, you know, which is actually what I did. I was, I sent an email just before our call and my agent helped me write it. And I'm cool with that. I read it. I know what the words were in it because they were all part of a 20 minute back and forth about this or that. And I'm like, hey, can you package this entire sequence up into an email? And it did. And I cut out most of it and edited it to be Adam. That's cool. You can even teach your agents to think like you think, you know, to like, especially in cloud with these plugins and stuff like that. And the way you can, you know, change the prompt behavior to make it more antagonistic to your ideas and not be like a yes agent.

1:17:32Yes. That was the most brilliant idea ever. Let's do this. No, no, no. Push back. You know, so you could do a lot of that with, with these things. And it's just. It's a, it's a bicycle. for your mind. With a mind. Yeah. And now we have it for our database. I think that's such a wild thing, man. But what's on the horizon for you? Like, I know that we can't uncover things that are maybe releasing soon or I'm not asking you to spill the beans, but what's over the horizon? A little bit that's coming up soon. Did you at least tease? What is that for you? Yeah. For Tiger Data, whatever. Well, so we started off as timescale building a database that scaled Postgres for time series workloads.

1:18:15And then it became a Postgres that scaled for workloads in general, especially analytical workloads. What we got into with Agentic Postgres is now serving not just the scale use case, but the start use case. It's like, hey, when I am building a new application with my agent, give me a database that just works. And what you'll see coming forward is just us continuing to innovate on those two dimensions, start and scale. Like on start, we're making our MCP even better and CLI even better. I'm curious what you think when these changes come out. We've actually been toying on a side project that is like a separate MCP server that is a little bit more than the database.

1:18:55And this is actually being developed by, I remember that engineer who said he would take a month and it took him 12 months. It's the same guy doing this one, but he's moving a lot faster. And, you know, I think we launched a new storage layer called Fluid Storage to enable kind of ephemeral forks. You'll see more stuff around that. Fast forking, safe forking, sandboxes, more around search and more at scale. Yeah. And I think, you know, again, like the job we do is, hey, when you're a developer and you're looking for a database, we give you the database that just works. and it's built on Postgres.

1:19:36You know, you can trust it. But just works means something different at start than it does at scale. At start, it means, hey, it just works with my agent, makes things easy. At scale, it means, yeah, it's also reliable and cost-effective and fast. Yeah, we're like 170 people and we're pushing on all those directions. That's a lot of people. I didn't expect that number. I should have asked that sooner, but yeah, That's a lot of folks. Wow. That's a lot of folks behind you. Pretty wild. Well, Jay, it's been a pleasure diving into where you've been and why you've built what you built and how you built what you built.

1:20:18It's been fun. Thank you. Thank you, Adam. Pleasure. Thanks for having me. Good to see you. Well, friends, there is something that stuck with me from this episode with a Jay. He said, build skills, not agents. And I keep coming back to that. You know, we're in this moment where everyone's racing to build something big, the next unlock in the workflow, the agent that does it all. Maybe, you know, skills are composable. Skills are teachable. Skills compound over time. And when you build a skill, you're not trying to replace someone. You're trying to empower them. That is a fundamentally different posture towards AI.

1:20:50And that's the thread running through my brain right now. That's the thread that I think Tiger Data is pulling on from time series to talking to your database, giving agentic access, letting you, the developer, be in charge. Talking to your database is kind of revolutionary. And on top of that, the terminal is evolving. The database is evolving. The way we ship software and build software is evolving, but the best tools have always done the same thing. They meet you where you are and they take you somewhere new. So a big thank you to our friends who have sponsored this podcast. and of course our friends over at Fly.

1:21:24That is the home of changelaw.com and maybe it could be your home too. Learn more at fly.io. And of course to the Beat Freak in Residence, Break Master Cylinder, bring in those banging beats and we love them. Hope you love them too. Well friends, this is one of the last episodes of this year. It's been an honor to host these podcasts for you. Hope you've enjoyed listening to them. Hang with us in Zulip. You can go to changelaw.com slash community. Hang your hat, call it your home. Everyone is welcome. No imposters at all whatsoever. You are welcome. Again, changelaw.com slash community. And it's free.

1:22:00That's it. The show's done. We'll see you in 2026. Bye, friends.

1:22:24Game on.