Why Agents Love Supabase — with Paul Copplestone and Ant Wilson

And I think the announcements are very much what I kind of look for in a company conference keynote. Um, so build is basically how they want to improve post- the Postgres experience for coding agents to set up. And the, the two that I, that really shout out to me is, uh, one, one is, is very simple one is, which is they're starting Supabase without Docker. That, that is like the, the less weight I can, I need in order to start up basic processes to interact with the, uh, the database, the better. Um, and obviously I can scale it up to, to a full VM in, in the future, but... Um, the second thing is Supabase Compute, where, uh, you can run Compute next to the database. I think co-locating that is super easy.

I think this is honestly super underrated. Um, I think is, uh, maybe underselling what it potentially could be. Um, your app having an MCP server, uh, from a Supabase library is really good and... But declarative schemas is so important, right? Like you want declarative schemas rather than the standard SQL paradigm of, uh, set up a schema and then migrate, migrate, migrate, uh, which is something we cover in the, the, the podcast. Like, um, if you, you want everything to be in Git as source of truth so that your agent can crawl it, um, and figure and set up and fork and what- whatever from there.

Um, it seems so simple, but really when Postgres was not designed to do this, uh, it, you have to kind of retrofit, uh, the schemas on top of, uh, onto a, you know, 40-year-old database. Um, and really I, I kind of see Supabase's momentum now as doing the last mile or like the, the really like frontier of Postgres development for agents to really use them super well. The beauty of it though is that it's not actually about agents. It's like we just needed this the whole time. Um, so but humans also benefit and I, I really like that. I don't like doing own, doing things only for agents to be happy. Um, so then the second part is operate- operating. Um, so that is mostly just about giving more context really, whether it's more logs, more checks, more connections, uh, markdown files for your SQL. [laughs] That's super cool. Uh, knowledge and skills.

They've-- every year they ship, uh, some interesting pipeline work from like, you know, Stripe to, to, to, to other sort of integrations. Um, and I think the sort of general ETL problem as you get further and further in your AI engineering career becomes, uh, you know, more and more part of the data engineering stack. But now Supabase has their own first party solution. You can use a mix of both, like whatever, but sometimes it's nice to have everything in one. Uh, the last part is, is perhaps the most, uh, complex and, and, you know, at scale thing. Uh, this is why Supabase changing their branding to Prototype to Petabyte.

Um, and really they're going after more and more scale obviously because they're grow- the customers are growing. Um, and so announcing a full V1 of Multigress, which is their, uh, their sort of port I guess of Vitess for, uh, for Postgres with the Vitess guy, which is, uh, enormous in the database world. If you don't know what Vitess is, uh, you're losing out. Um, you, you need to know. You're, you're watching this on YouTube, uh, on a Vitess website. Um, OrioleDB and DB... So like there's a lot to get in there and honestly, uh, you can see my connection with the, with the founders. I think this is the first interview ever that the two of them have done, and I make reference to that and try to, uh, view their chemistry and show, show you what it's like to discuss and run a $10 billion database company.

Enjoy. [upbeat music] Okay. This podcast has been six years in the making. [laughs] We're excited to have, uh, Kapul and Ant in the studio. Welcome guys. Thanks. Thanks for having us. Uh, we're recording this ahead of Supabase Select, which, you know, it's your, it's your big customer conference, and we'll release this after the, the conference. But I just wanted to sort of celebrate the journey that you guys have been on. I've seen you guys since, uh- You guys finished YC, and you were, like, in that place in Singapore where you were sort of celebrating things. And I, you know, I, I met you guys for the first time, and I was like, "What is this, like, hot database company that, like, got s- like, soft launch on Hacker News by, like, you know, some random user or something?"

Um, anyway, congrats on all the, the, the success. What are you announcing the headlines at Select? Let's just get right into that. Yeah. Yeah, sure. Um, so there's quite a few headlines. We've got a lot to cover. The main ones would be starting off with, um, Compute. So we offer a product called Edge Functions, and Edge Functions are powered by Deno, and they have basically a two-second limit. What we're finding is that... Well, what we've heard from our community for a long time is that they wanna run much longer workloads. They wanna do things like chunking embe- embedding. So tomorrow we're announcing, and by the time this comes out, Supabase Compute, which is you can run anything. It's long-running. It's serverless.

It is running in the same private network as your database, so it's co-located, has very good latency, so you can run Postgres connections directly from Compute. And you can run anything inside it, so it's no longer constrained to just Deno. You can use it with Rust or Python. You can run agents inside it. It's basically just a, a full compute instance. Let's call it, uh, the comparable offerings in other clouds that others might be familiar with, uh, would this be- Yeah. So something like a, a Lambda function, but, um, the container type Lambda function. Okay. Yeah. Okay. You can run a full container, basically. And then there's the other stuff. Yeah. And then we have, well, three dem- uh, innovations on the database space.

So the first one, which we talked about last year, which was OrioleDB. OrioleDB, we've basically put it through the ringer [chuckles] in the past year to make sure that it is rock solid and ready for the Postgres ecosystem. The three innovations are basically, uh, performance, scaling up, and scaling out. Uh, Postgres touches on... Uh, sorry. OrioleDB touches on performance. It is a drop-in replacement for the c- the current Postgres storage engine, which is Heap, and it runs about a hun- 1.5 to 2 times faster, and you don't have to change anything else. It's just completely compatible. It completely removes the need for some of these background tasks, like vacuum and bloat, so it's actually very efficient for your database as well.

And basically, the innovation that all the work that we had to do this year was put it through pgregress, full compatibility with Postgres, antithesis. We put it through TPC-C nightly benchmarks and Jepson to make sure that it basically can never lose data. So we're releasing it tomorrow in full beta, and we also made it Apache 2 licensed. It comes with a patent, and the Apache 2 license covers the patent. So we actually acquired the technology, continued to develop it, and then we paid for the patent and are now giving it away for free. One thing that you mentioned in passing, which I still don't... I can't get over, is that you tested every operator and you said, um, every operator drops data in some situation. Well, so this is for the second- Okay...

We've got a whole design partner program, so, um, we're working with the likes of Booz Allen, we're working with Postman, we're working with Perplexity. So yeah, lots of customers coming on board. They-- A lot of people need it. They're running Postgres at scale, and they need something to sc- help them scale up. Yeah. Third. Third one is, uh, yeah, so this one's a big one. Um, scale out.

And it's really when some of these innovations came along that we realized, "Oh, okay, we can do this." You know, you, you're going to five million a month. Let's say you see on the, the trajectory of b- billions per month. One of the secrets is they're actually kind of small databases. Yeah. Right? Like, these, these things are tiny. They're vibe coded apps. Like- Yeah...

n- like, I, I, I don't know if you have any insights on, on the, on, like, what features were- Yeah. I mean, until they get traction, right? And that's the thing is, like, once you hit the front page of Hacker News, you need to know that you can go far beyond that. Um, and so the whole idea with, with all of this was we need to make sure we've got a really smooth pathway to migrate between the back ends, like zero downtime migrations. It's like we, it's just, like, a, a need to have. Because, like, we see now that there's a lot of non-technical builders who are building, like, whole companies, right?

We have the, a customer, Magic School, and the guy was, like, a teacher, right? And he saw all these, like, uh, inefficiencies in teaching, like teachers spending time that, on, on tasks that could be automated. And so now you have these startups that start without a technical co-founder, without a CTO, and that's not, like, isolated incidents anymore. That's, like, becoming more- Everyone... everyone's- Yeah... like, a bu- a builder now. Um, and that's why we see, like, the growth we've seen over the last 18 months is crazy. It's nuts. Yeah. But, but it still feels, like, early. And I think, um, there's a much larger market of almost, like, consumer, like, builders, right? So people- Mm...

inside companies, people who wanna launch their own startups, which is gonna mean a lot more databases, which- [laughs] Like we said, it's like if we give everyone a dedicated Postgres cluster, it's probably n- you know We'll probably run out of physical computers in the, in the world pretty fast Which is basically what's happening [laughs] Yeah [laughs] Yeah. I mean, like, we've faced it throughout this year. Yeah We just, our growth has, uh, has been, uh, one of the things that we've had to deal with in terms of capacity of computing. Yeah. So congrats. Also, I will observe that I feel like Glauber is also a meme lord. Yeah. Well, that's one of the things- Right?... why we like him [laughs] I mean, like, come on. Yeah, he's amazing.

So for those who don't know, the, the person behind the Supabase memes is this guy. Yeah [laughs] Uh, that, you know, um, low key, low key a killer- [laughs]... I would, I would say. And, well, Glauber also. Yeah, yeah, yeah. Big time. Uh, and he- I don't know if you guys ex- exchange notes on, like, what's funny and- Not yet, no... things like that. No. We're gonna give him- Oh, that never came up? We're gonna give him access to the Supabase Twitter account- [laughs]... for sure. Oh, my God I could, I could do with some help [laughs] Uh, yeah, yeah. Wait, is it still you? Uh, I... We've got a small- Okay [laughs]... team of, uh, teenagers now who help me out. [laughs] You do workshops, right?

Every week Yeah, but it, it's ha- We do a weekly workshop, a meme, basically weekly meme workshop, but it's hard to hire for because you need to understand the tech and understand, like, what developers find funny, basically, in order to do that type of marketing, right? Yeah. Um, ac- Actually, you're, you are also low-key- Yeah... like, meme lord. I'm not as funny as you. You're really, really funny. Where's... There was one with, like, a seagull, and you were s- like, um- Ah. [laughs] Yeah... standing on top of the- Supabase... turtle Standing on top of Postgres, I think it was- Yeah... with a seagull Yeah, it's like a standing seagull riding a flying seagull, and it's like- There we are.

Oh, my God. Supabase Postgres Yeah. I, I can't, I can't find it right now. Yeah, yeah But, and this is all, like, deep Supabase history. [laughs] But like, you know, there, there was a time when, like, those things were just like, it was so good, people wanted to be like you, and they couldn't, and that was your moat. [laughs] Like, memes. Yeah. And there were some big companies trying to run memes off their Twitter accounts. I won't name any names, but, like, very large organizations. Um, but yeah, it's, it's hard. Like I said, it's hard to hire for- The guy who did, like, a 40,000-word essay on breaking down our meme strategy Mm Someone studied our Twitter strategy, actually really early on, and posted it on, on Hacker News.

Yeah, on Hacker News. Uh, and, and that was funny because- [laughs]... they'd recorded the strategy in a way that we hadn't committed to paper yet. So then when we onboarded new marketing people, we were like, "Here's the strategy." [laughs] That was written by someone else. Yeah. And of course, Hacker News hated it, and then they were commenting on it, and- I found it. I found it. I found it... Ant took the, uh, the first comment This guy, this guy, this guy. Yeah, yeah "Learnings from four months of Supabase feed." He just looked at your work for four months. When was this? 2022. Okay. Yeah, yeah, yeah. Wait. I m- I, I mentioned it? Ooh, okay.

There you are So he just broke it down He broke it down. [laughs] And then Hacker News hated it, on it, and so Ant took a photo of them shitting on it and posted it on our Twitter account. [laughs] [laughs] Yeah Your hate just fuels me. What is this? Uh, interesting article. Yeah, okay, so I commented. [laughs] Yeah. Yeah That was like, a lot of our growth was thanks to Hacker News. I mean, i- in the early days, I just- Yeah... had to live on there and- So we should talk about that. What deal with the devil do you have to do with Dang, uh- Oh, man... to- We, we've been scolded- [laughs] Yeah...

by Dan G, like, many times- We have... like, for l- for launching too many things in a week- Yeah... basically. And he's, like, rightfully paranoid about the association of Hacker News- Yeah... and YC. Yeah So he kind of like, if something happens, he kind of wants to throw us under the bus so that- Yes... he, people know, oh, we're not doing this for YC customers. We're an independent site. So he's been helpful to, uh, make sure that we, we don't smear the name of, of YC- YC... or Hacker News. I mean, at this point you don't need them. But, like, for sure, I think, uh, Dang does a... Dang has super high credibility. Yeah No one, no one really, like, you know, uh, questions him.

But I, I do think, like, you know, you guys innovated the strategy of launch weeks, and for a while, I think there was one of the launch weeks, I think four or five, where every single launch was n- like, number one or page one on Hacker News. I got told off so badly for that because I- But what-... tweeted that proudly. I was like, "Oh-" Yeah... "here we are." So what? And then I got told off 'cause, uh, we were, like, it basically looks bad because- [laughs] Yeah... because, like, suddenly this YC company is, like, dominating the front page of Hacker News. It looks like it's a biased thing. And, like, parading the fact- Parading... that you're on the front page. Yeah, yeah, yeah. It's just great. Yeah, yeah Which is true.

Like, so, like, tip to anyone: Never screenshot the fact that you're on the front page- Yes... of Hacker News and put it on Twitter because it's, it does look like we're... And, and to be honest, it, it was gloating, so, so it's a fair, fair call-out. Yeah. Yeah. There's all that, which, like, I would say, I would cons- consider that somewhat ancient history now. You know- Yeah... it's been a few years since, since this was a relevant thing. But for those who don't know Supabase history, these are the things that you must check off as, like- [laughs] Yeah... the, the tour of things. Sure. What else? I mean- I, I think that the, the top post there is interesting because- This is the launch post... yeah, but, but it was someone else who launched us. Yeah. Vicky. Like, we weren't ready. Yeah, yeah, yeah.

So it was, like, totally fair at the time, just given the, um, the way the world was looking. People made many irrational th- uh, decisions during COVID. Yeah. But you never know. I was gonna say, uh, y- you might be worth more in DigitalOcean now, but then somehow DigitalOcean has also, like, gone on a super tear. Ah, cool. Look at this. Because computers- True, true, true. They tripled, they tripled in the last year. Wow, that's cool. Wow. So what are they, 16- They're 16 billion now... billion, yeah. Epic. Okay. Uh, and like, and like, uh, okay, well, I guess there's a shortage of CPUs. Yeah, there's a shortage. [laughs] How are you guys experiencing it? [laughs] We literally... I mean, we're up against runways all the time where, like, we meet AWS, like, multiple times.

Is there a way to spike when you have a spike? Or what do you do? Like, you, you just have agreements with people to- Yeah, we've gotta, like, plan- You just pay more... very closely. Yeah. We, we pay more, we partner with them. And in fact, tomorrow as well, we're announcing that we're bringing on a new investment partner, which is CapitalG. CapitalG basically has access to the Alphabet ecosystem, good for us on both sides. Like, we now have to consider a multi-cloud strategy. Meaning GCP. We need GCP, or we need more compute, basically. So GCP is a very obvious choice, and of course, we wanna partner with all the AI labs as much as possible. And so they have a suite of products.

We work with them already on Gemini, and we wanna make sure that we're embedded deeply into all of their products. And that's another thing where you're mentioning how so much of the database ecosystem is being shaped by block storage and all these innovations and all these things. Clearly, AWS is, is, is the leader there. Mm-hmm. It's hard to be multi-cloud if you're gonna use GCS as, like, a swap for S3. Yes, as long as they've got the features. Actually, I don't know if their object storage has the, all the features, but I'm sure that, like, whatever S3 do, others follow, so... [laughs] I re- I still remember, I think, like, three years ago, they introduced, like, strong consistency.

Yeah, that's the innovation basically that- Right... that helps. Yeah. And they just, like, they did it for free, and then- Yeah... just rolled it out to everybody- Yeah, yeah... and it's like it's live now. Yeah. What? [laughs] Yeah, yeah. And then S3 Express as well, that was a big game changer, um, because- Why? Well, basically now what you wanna do is, like, you typically write to a local drive, and then you need to batch it to something more durable. And that batching process, um... And so the write times to this are one to two milliseconds. And so before you can guarantee a write back to your client, you basically need to batch it to something that's durable.

So this one's durable, not as durable as S3, not 11 nines, but durable- Good enough... enough that you can f- It's ef- ephemeral durable. Yes, exactly. And then, then in the background, you take, you take-- you use this kind of as your cache, and then you move into S3 in the background, which is cheaper. So you don't-- do we need Redis anymore? Uh, yes, yeah, you still need Redis. [laughs] Like, this is no, nowhere near fast enough. You need, like, basically, your cache layers are either memory, NVMe drive, um- Interesting... locally attached to S3, their equivalent, and then S3. Yeah. Normal. Amazing, amazing. That was a little bit of databa-- Like, we, we jumped around quite a lot- Sure. [laughs]... uh, clear business memory.

Uh, a few more things that we're gonna talk about. You, you... your slogan was build in a weekend, scale to millions, now we're prototype to petabyte. Just-- Which is excellent. And really it is reflective of the evolution in scale that you guys are doing, right? Yeah. Do you have anecdotes on, like, let's say, early Supabase versus now? How fast are people scaling when they, when they start a thing? Yeah. Well, the big thing is, um, I think how fast people are monetizing. Oh, yeah. At, at the start it was like you choose your database, you build your product. At some point, you, um, add in billing.

Now, because you can add in Stripe, like, at the agentic layer, it's like we see customers starting a project and within the same week, like, generating serious cash. Yeah. And, and in the early days of vibe coding, it wasn't obvious to us that this is something that's gonna stick around, this is gonna change the way software is built. But as soon as we saw someone earn, um, like, I think it was $10,000 within the first few days of launching their product, we were like, "Wow, this is, this is real." Yeah. People are making real things that generate real money. Um, so we knew we had to lean in, in, in a way where there was waves we'd ridden successfully in the past, like when NFTs were huge, we leaned in a bit and we added, like, sign in with Ethereum and sign in with Solana, um, but we didn't dedicate, like- Tons of dev time because everyone kinda knew it was gonna go away at some point.

And making that decision for, for vibe coding was like, you know, we, we discussed it a lot as a team of how much do we lean in? And it was a time when we knew we needed to go slightly more upmarket. Like, our biggest customers were, like, stretching the limits of, of what we had at the time. And so we knew we had to then just do both. It was like, you know, um, innovate on the, on the small, like, prototype stuff and innovate on the, on the much larger side as well. And then you just need more people for that, so, [chuckles] right. That one's funny because, like, on the numbers side-- B-basically, for Supabase, there's been three chapters of AI. So the first chapter was, um, embeddings. I think we were on- Supabase Vector. Yeah.

Which I have now transcribed, and it will always, uh... We, we, we should probably update this 'cause, uh- Yeah, yeah. It's probably super- This is the only talk you've ever done at AI. [laughs] Yeah. [laughs] We did this talk. Uh, basically, all these embedding databases were coming out, vector databases, create the vector. And so we were pushing the fact, and you can use Postgres as a vector database, and Andrew Kane, a, a completely independent guy, has developed this awesome extension. And we're like, "Okay," trying to educate the market. Postgres can kinda do anything. You don't need another database if you're gonna store your own embedding.

So this was chapter one, and it was, it was really a, a good growth driver for us, uh, and also good for the Postgres ecosystem, just letting people know. Now there's a lot of innovations in, in embeddings. What Turbopuffer, for example, are doing, they use a lot of the innovations that Turso are doing, and I think it's a phenomenal use case for that type of stuff. So that was chapter one. Chapter two was, like, the vibe coding wave, whereas Lovable, Bold, all of these ones coming along. That was end of twenty twenty-four, and that's where we debated a lot. "Hey, do we really wanna be there? Do we wanna be servicing these? Do we even wanna have our own vibe coding type platform on top of Supabase?" But in the end, like, our mission is store the world's data.

And to do that, we, based on our tagline, it's like, be there when you're getting started, make sure that they never wanna leave. And so we saw people were getting started. They were doing toy things, but still, it fit our mission. We just had to believe that over time, these tools would become better. And then the third one, which is the, like, crazy numbers, start of this year, the models just got better, basically. And then we started seeing our top of funnel, our, our sign-up rate start rapidly increasing. But usually when this happens, you're pouring a lot of dollars into marketing. Mm. We actually barely spend any dollars on marketing, and, uh, most of it's just memes and things like that.

[laughs] So, um, so- Here, I got, I got one of your charts. There we go. Yeah. I mean, all, your, all your charts look the same, but because they're just, like- [laughs]...exponentially keep scaling- Super weird...you just keep, you just keep- But-... changing the Y-axis... the other crazy thing about this is, is often when you get, like, more consumer life traffic in the top of funnel, your activation and conversion to pay goes down. Yeah. But both of those things went up. Started going up quite rapidly as well. Yeah. Yeah. Because of the model. How do you know activation and conversion to pay? Oh, your-- at Supabase's. Our, our metrics, yeah. Ah. Our funnel metrics. Okay.

Um, and, and the reason is because the sign-up to activation, the agent does it for you- Yeah... like in one step, so it, it's basically a hun- it's almost a hundred percent- The friction's gone... of people who convert. Yeah. Um, and so those users then were converting to pay at a higher rate than our traditional developer user. Okay. I mean, excellent. It's great. Yeah. Great. Thanks. [laughs] Good. Uh- But, but I'm sure, but I'm sure, I'm sure, like, like all developer tools are seeing the same- Yeah... thing, right? Well, at, at different rates. I think you guys are clearly one of the, the leaders there.

You know, I was gonna bring out this, one of these, these things. I have this AEO tracker, uh, where, you know, if you, if you type in Supabase and there's a, there's a few categories where you're the leader. Uh, but basically it l- has led to every single frontier model recommending you as your primary choice for an integrated backend. What does integrated backend mean? It, it is your original offerings, like your, your auth, your, uh, functions, your, you know, um, and your database and, and all the other stuff, uh, APIs. And I think it is interesting, like you had to be in the market for three to four years, exactly your- Yeah...

time zone before LLMs took off, and then present yourself as the leading competitor to the leading thing at the time. Yeah. And then to-- you can take over. Yeah. Yeah, yeah. [laughs] We're very lucky with our timing. [laughs] I mean, we couldn't have timed it better. Like it was a complete accident. Well, I mean, I think like on first princil-principles basis, so we have interviewed other founders who maybe started post-LLMs, and Resend was kind of their... I mean, we've also talked to them, like, "Why does my LLM keep recommending Resend?" And it's because they have a, a lot of good AEO practices. Sure, yeah. And of course, we invest very heavily in that.

But, you know, when-- w-we've been at that point since the start, and that was an accident, right? So now we just need to maintain that lead. And community as well, right? Like our, one of our growth strategies early was just, was to spend to- so much time with the community, and it resulted in they were writing blog posts, they were making YouTube videos, they were putting out podcasts all about how to use Supabase, in what situations. Like, "Here's how I built this thing." I'm sure that all just got slurped up by the- Yeah... the models. And open source helps as well, of course. Yeah, yeah, yeah. Definitely. Being, being open source.

Uh, you actually had a fun announcement on the community side because you're raising so much, and you're trying to involve your community in some, in, in, more deeply. Yeah.

It can be confusing because there's Gith- uh, there's Supabase, the open source- Mm-hmm... and then there's Supa- Supabase, the cloud platform. Mm-hmm. And s- and most people are probably gonna come into support for Supabase the cloud platform. Yep. So yeah, it is. Yeah. Do you have any idea of your, uh, uh, uh... This is very hard to tell. Your open source market capture. So, uh, this is a thing that, like, if you're in the developer tools circles, you, you know. You do? We, like, quite deliberately do not- Don't know... so- Yeah... we don't put any tracking in our open source offering. You can use it.

And we've had enterprises, large pharmaceutical companies, for example, come to us and say, "Hey, we're running 100 instances of your open source project. Can you offer us a product?" So we've got this product called Control Tower, which allows enterprises to have-- monitor a fleet of databases. You won't find it on, on our website, but it- Oh, come on. Yeah. [laughs] Why not? 'Cause it launches- [laughs] Oh... tomorrow. Tomorrow. [laughs] Oh, okay. Okay. I don't know if we're announcing it tomorrow, but yeah, basically, it's Control To- it-- All it does is, it's this fleet view. We've got this very popular product, Supabase, for platforms, so that's what all these AI companies are building on top of. We launch millions of databases for them.

And so all we needed to do was repackage this for enterprises, and enterprises as well want to do the same thing, which is they wanna spin up many- Internal Supabase... many internal Supabase for all of their users. They wanna see and administer them. And so Control Tower is essentially a repackaging of what we've already been developing for these platform pro- providers, except specifically for, for individual enterprises. Yeah, I do wanna cover, because I think a lot of people think of you as bottoms up GTM, uh, you know, what the work you've been doing on the, on the enterprise side. Mm-hmm. Uh, so first of all, shout out to Wenbo. Yeah.

[laughs] I, I think this is your most successful recent launch, which is- Uh, Supabase for platforms. Yeah. This one launched in... Yeah. Yeah, yeah. I mean, December 2025, so yeah. Uh, yeah, but- I saw some stats or, like- Yeah, yeah... you, you guys call this successful, I don't know. Yeah. Um, what is the enterprise side, right?

Uh, interestingly now we have like CPOs who are like, "Wow, I didn't code for the last ten years, but now because of Lovable and Supabase, I'm building all these things internally," that we now have like champions at the, at the C level and without needing to put like a huge motion around that.

Oh, that's great. Yeah. Mentioning Lovable, mentioning Bolt. Lovable is gonna have their own database stuff, you know, and, and like everyone is gonna basically try to intrude on, on the, each other's turf, uh, which is normal in, in the, in the grand scheme of things. Which is more sticky? Which is more sticky? More s- more sticky. Like is the CPO more likely to be married to Lovable or married to Supabase? You know, like- Ah, okay... is, is that a question that comes up or? That's a great, that's a great question. I think ultimately running databases is very hard. Fair enough. [laughs] Well, and, and as we just discussed, it's harder to migrate your database than it is your- Yeah, yeah...

front-end tool. Yeah. Um- Yeah... not that I don't think Lovable is sticky. I think it's amazing. And actually, they were the first people to do a really good job of making sure their agent didn't do stupid shit with, um, the database. Because we had, uh, very early on the agents were like trying to shove image data into Postgres. You know, they were turning off RLS in order to get [laughs] their projects to work. Hmm. Um, by paperclip maxing basically, is that Lovable wrote really good skills to stop their agents doing stupid stuff. They were, they were the first. That's the secret sauce. Yeah, yeah. Yeah.

Is to, to solve all these- And, and, and now like the models themselves don't do that stupid stuff anymore. Yeah. So it's, it's, the situation has improved a lot. It's gone from like, um, Will Smith eating spaghetti circa, you know, 2022 to Will Smith eating spaghetti in 2026. Yeah. Yeah. Yeah. Um, and that's a, that's a famous meme. [laughs] Uh, talking about... Uh, I'll, I'll leave the migration point alone after I mention this. You know, one of the things that I got like a big aha moment was when I met Karen from Co2- Yeah... who's one of your board members, and she said, like her thesis is whole, like, you know, you are not just, uh, replacing Firebase, you're also like, you know, coming off as like new, new RDS, right? For- Mm. For Amazon.

Um, if there was a, you know, stage four of, or fourth era post-AI of Supabase that is like the other, you know, the one trillion X in your, in your growth, rip out Oracle. I, I don't know. Like- Mm-hmm. Yeah... what, is, is, is that the- Well, that's the, that's the- Is that what the white whale there? That's the one that the famous migrations are after.

This is just my down database indulgence. We also had two of the co-founders of Databricks here, Matei and Reynolds, where they are making a claim that they have solved HTAP. Okay. Yeah. Which is the sort of the white whale o- of database people, right? Yeah, yeah, yeah. If you, if you know, like, uh... Well, traditionally you need an OLTP, you need all that database. Yeah. Um, never the two shall meet, except that, uh, sometimes you, you would like to never have to replicate anything. Yeah. How you... Are you paying attention at all to that stuff? Like, is, is there, is there a story there for Supabase some- someday? Yes, we are particularly interested in Duck Lake. I think Duck Lake is a very... and DuckDB is a very interesting technology.

Is, is it Mother Duck, but a, but a data lake? Like, what, what is, what is Duck Lake? Uh, Duck Lake. Okay, so, um- [laughs]... DuckDB- Yeah... which was recently... It's DuckDB, which is basically like a SQLite for analytics. Columnar storage. Yes, exactly. Exactly. It can read directly from object storage. That's typically how it's touted. And it can actually read from your Postgres database as well, so open source, great technology. And actually there's also, uh, DuckDB Labs, which was recently acquired by AWS. Yeah. So they're a foundation- [indistinct]... that are developing. Yeah. So they're building a committee, um, around the open source.

I actually think of DuckDB a little bit like the Postgres of the analytics ecosystem in that they're trying to make it so that it's owned by no one. It does a lot of things. It's very versatile. It's like a professor in Netherlands? Yeah, exactly. They live, they live and work out of the Netherlands, which is pretty cool. So our interest mostly on the warehouse side is not trying to make Postgres an analytical database, but mostly then focusing on using the right technology for the right job. I don't know if HTAP... Many people have tried to slay the HTAP dragon. Object storage could be the solution to that, but generally we think once you get to a certain size on the OLTP side, you need to have sharding.

There's a limit to the amount of NVMe disk space that you can have, so if you need to split transaction across two servers, you're basically then going to be fetching it from object storage. Eventually, that will slow down your database. So we think that sharding is by far the best way to do scale up into the petabytes, and then you should treat your analytics differently. Now, that doesn't mean that you shouldn't store your data directly on object storage, for example, and then read from it from a different route. But trying to smash these two together- Hmm... doesn't necessarily make that much sense in my opinion. Yeah. I, I... I'm happy to leave it there because we do have more, uh, to, uh, work on.

I do think, you know, they, they have some innovations where they're, they're basically using their s- their spare compute on their, uh, database, uh, machines to power all this, uh, replication is, is my read- Uh, yeah... of what they're, of what they're doing. And actually, to be honest, I haven't looked so much into how they're doing HTAP, so I could be misquoting. Yeah, e- exactly. Yeah. It's, it's like you're, you're in your lane. You know, it's not exactly like your customers are, like, banging the table asking you for this. Yeah. It is just, like, if you want to become the preeminent database company, someday you have to have an answer. Yeah.

[laughs] Yeah. Exactly. Yeah. Which, like... Yeah, it's, it's interesting. Uh, the other mention I will check in on is durable workflows and, and, you know, that world- Uh, yeah... which, which as you know, Temporal, uh- Yeah, yeah... I, I've been, I've been, uh, banging the drum for a while. You have made, uh... Like, every year you make... you do a little thing. Yeah, yeah. We're incrementing our way towards [laughs] dur- Yeah. But, like, don't you... Like, how come it's not, like, more P zero for you? Because, like, you know Temporal is doing super well, right? Okay, so what was missing for us to do durable workflows? Compute. Right. So you're, you're doing this now. [laughs] Yeah. Super interesting space to be in, I think. Like, yeah, now we have had people in the community try to build workflow engines on top of Supabase functions, and it just didn't work because of the two-second timeout. Yeah.

Now we've got the primitive that would unlock the ability to do long-running durable workflows. I think my original pitch for Supabase is, like, you want the sort of the all-in-one, the, the, the sort of agentic cloud- Yeah... that, like, is the backend, integrated backend that, that people can, can use, right? Like, you've got sign-in with ChatGPT now, which is, uh, a great part. Like, you, you just kind of have to sort of give the AI everything it wants, and then it will be your best friend. [laughs] Yeah. Yeah, pretty much. [laughs] Pretty much. And be prolific. I mean, our P1 strategy is to partner with as many of these frontier labs and models as we can, so we've gotta be prolific there, and then we basically, uh, we see our attach rates just going up as well. Yeah.

I mean, obviously, there's the US labs, and you're, you're, you're here in the US. Do the other labs have significant partnerships? What, what are they like? Mistral- Ah... Cohere, the Chinese labs. I, I don't know if- Not so much... what's in your- Well, well, they might- Okay... they might do. Uh- Okay, you primarily care about the US labs. Yeah, yeah, yeah. [laughs] Yeah, okay. Okay, that's fair enough. [laughs] Yeah. Yeah. Mainly just 'cause, like, that's who's delivering the most traffic- Mm... at the moment, right? That could change. Yeah. But what does partnership look like? Like, basically it's one thing to be on the, their app store. It's the other thing to be their default thing when they- Sure... when they launch an artifact. Yeah.

So that one's super interesting for us to be able to hopefully be one of those providers. Mm. One of the reasons why we couldn't in the past was because we just couldn't handle their scale. I mean, take ChatGPT, for example. What have they got now? A billion monthly active- 1.2... 1.2 billion monthly- Yeah... active users. One of their users launching a database every single month. [laughs] Now we've gotta launch 1.2 billion Postgres servers. It's just, as I said, we've gotta get to the billions of databases per month. We just haven't been able to do it until today, where we acquire a company that means we've got the technology which can drive it up. Scale up, scale out. Okay.

Um, a- other than that, what are other, I guess, P0 objectives for the company? Like, you know, over the, over the next year, how are you sort of guiding the, the direction? Our P0s are basically win the strategic labs, win the models, and, uh, continue our PLG push. Where we are focused on the enterprises is winning these- Yeah... new workloads. And it's more that, like, the ceiling of PLG, when we started the company we were like, "We can get to 100 million on PLG. We can probably get to 200 million, but then we're gonna need to do something different in terms of motion." But because of AI, like, that ceiling just lifted, and we don't know where it got [chuckles] lifted to.

Like, is it a billion? Is it two billion? Is it... You know, you could just-- we could just keep going. But we still wanna do the other stuff as well, so we basically, like, last year we were like, "Well, we have to do both." Um- Yeah. [laughs]... the bo- the small stuff and the big stuff, and, you know, uh, maybe we're a bit greedy. [laughs] No, I, I, I think it's, it's what people want, right? Yeah. The, the same sort of mental model that, uh... And, you know, they'll, they'll pay you to handle the fine grain details of, uh, moving up in terms of orders of magnitude. Yeah. And I don't know how many orders of magnitude that is, but quite a bit, quite a lot.

It's just crazy, because if you would've asked us 18 months ago, we were growing fast. If you would've said, "What can you do to 10X your top of funnel?" I would've said there's nothing. Like, it's, it's crazy, right? And then just this whole market just exploded, so let's see what happens. Yeah, I remember at our company offsite at the start of the year, we, we put this crazy ambitious revenue number out, and someone came up to me and said, "Oh, you shouldn't have said that number. Like, the team are gonna be so-" We're gonna miss it... we're gonna miss it. Yeah. The team are gonna be so disappointed, and it took us, like, four months to hit the number.

It's just, like, a ridiculous growth curve from the start of this year. Amazing. Um, what are the open questions that you have? Like, you know, if you had a crystal ball, it, that it would change the way that you bet on things. How development will be done- Okay... and what people will develop. These are the two things that I think a lot about. Okay. You know, what are the possible answers? It's just more, more agents. Like- Yeah, so, like, this year we're focusing a lot, like, and I'll talk about it in my keynote, basically, that we're moving from... Like, we started as a dashboard company. That was the best DX. Then, like, you know, a lot of people are like, "Oh, headless mode."

Headless was what we kind of started, Headless 360, Salesforce, all these sort of things. I mean, we had that at the start, like, you could use Supabase directly, API-driven. That's one of the reasons why we succeeded. This year, our big focus is moving from, from, um, to code first, Git native. So basically, when you launch a Supabase instance inside, um, your local machine, you get this Supabase folder, and it represents your database. It has all your functions, everything. Does this exist today?

you have strong opinions on, like, how the software should be built, right? Like, in terms of, like, uh, continuous integration or, like, continuous delivery or whatever, and I find myself, like, pushing my opinions on the, on the agents of, like, "Oh, you wanna do it in this way," and...

But now more and more, like, you don't need to. But my worry there is, are they all just gonna, like, converge on one way of building software or- Yeah, mode collapse. You know- Which is great. They're all gonna use Supabases. Fantastic. [laughs] Yeah. But, but, but, like, the g- the thing that's great about, like, the software development injury is, like, i- industry is, like, these are, like, famous flame wars, right? Of, like- Yeah... what's the correct approach. Like, are we gonna lose that or- Yeah, so you want diversity in ideas- Yeah, exactly... so that the best idea can win, and but then the models are just gonna make you so lazy that nobody even bothers to look at the details.

Yeah, and, and they just all converge on one way of building software, and then that's, you know... Is it gonna be like a local maxima in terms of- Yeah. I, I don't know. I, I, I do think, uh, yeah. So first of all, I think that's a valid fear. I do think it's out of your control. [laughs] Um, I, and, and, yes, uh, the low taste, let's call it the casual app builders will, will converge on whatever the, the mean is, uh, the, the model is, which is you guys. But then the, the discerning high-skill people will always really care about the details as, as they should. Yeah. Right? And which- Yeah... which will, which will then filter down eventually- Yeah... is the, is my idea.

[chuckles] But are these high-skill people gonna continue to be trained by- Uh, I don't know. [laughs] Yeah, yeah. I mean, I think, I think people are passionate enough. Yeah. Uh, I, I... It's one of those things where, like, you know, I do think, like, uh, you know, the average, um, college is probably screwed, right? Like, they, you probably don't need to go to university anymore because you can probably self-teach and just, like, you know, uh, get, get your own, uh, job. It's always been, like, this top-tier people that, that are moving a lot of the, the, the needle, and you need to, uh... And, and they will always find a way to have their own self-interests, uh, be satisfied.

Curiosity in terms of, like, what, what is possibly better. You know, the type of people that take a year off and go to the Recurse Center and just work on cool programming problems even though they could be making money in, in industry- Yeah, definitely... and, like, yeah. [laughs] Definitely. [laughs] Yeah. What do you disagree on between the two of you? It's rare to have the two of you have, uh- Wow. Oh, yeah. This is great... on, on the pod, so, you know, I wanna capture this. I think one of the reasons we've been so effective, like, working together is because there are definitely situations when we'll disagree, but as soon as the other person, like, pushes a little bit- Yeah...

the other person's like, "Yeah, maybe you're right. Let's, let's try it." Yeah. Um, so we never end up in, like, big bust-ups, um, and it turns out, like, most of that stuff actually doesn't matter, right? Yeah, it doesn't matter. [laughs] Anyway. Like, as long as the general direction is, is correct, then, then it's gone okay. But, uh, yeah, I can't think of- Oh, you guys are really struggling here. Oh my God. [laughs] Yeah, yeah. Which is why, like... Because it's, it's never an important enough thing. Like, we've never had a bust-up because it's never been important enough to actually fight hard enough on it. Like...

And also, we're fucking busy, so, like- [laughs]... usually, like- You got your thing, I got my thing... Ed's handling... Yeah. Yeah, yeah. We're trying to handle different parts of the business. Yeah. Maybe it's just, like, things are growing, and obviously you got your domain and, and you just need to work together, right? Yeah. Yeah. Yeah. Yeah, it's interesting, you know. Uh, so Laura here, uh, all three of us were in Entrepreneur First- Yeah. [laughs] Yeah... in Singapore. We were all different work- working on different companies, and then you guys stayed back after the cohort? Uh, you were housemates after the cohort? Yeah, yeah. Yeah. So we were all-- We lived in a house with seven people, and we were all doing different startups. Yeah. And so Em was doing his startup.

I was doing another one. This is before Supabase. And so, like, I think somehow your energies, uh, you know, you, you, you found that you were suitable working together and, and working on this, uh, this, this project. Yeah, because, like, traditionally, especially in, in an accelerator model, you're pushed as a technical person to work with some domain ex- domain expert, right? Um, but in the house we lived in, it was basically seven people who were all builders, all technical people, and we found that pairing to just be, like, so much more effective, especially on when you're building dev tools 'cause you are the subject matter expert as a developer, is that we could, could just get so much more done.

And so then we ended up hiring most of the other people who lived in the- Yeah... the flat, as well as our early employee- Rory... is Rory, who- Was the first employee? Um, not the first, but the first e- to, on go-to-market. Like, now he runs- Yeah... all of go-to-market. Uh, the Postgres guy was first. No, uh, Angelico. Angelico. Yeah. Yeah, yeah. We had an intern- An intern from my previous startup. Ah. We just dragged him along. [laughs] Yeah. And, and then the other, like, first, like, full-time hire was this guy Steve Chavez, who, um- Postgres, yeah... who was the Postgres maintainer. And the funny thing is we were in the YC batch, and the advice they give you in YC is like, "Okay, we're gonna give you a bunch of money. Don't hire anyone."

Uh, that's the big temptation is to hire a big team. And we went to our partner, and we said, "Oh, but we've got this guy Steve, he's the lead maintainer of Postgrest," and he was like, "Hire him immediately." [laughs] You know? [laughs] He was like, "Forget the advice. Go and get him." W- of his reputation or because it made sense for Supabase? Well, yeah, both. It, it just made sense that, you know, this was a technology we were very reliant on- Yeah... and we were gonna lean in, and it was like... And the, the funny thing is, we hired Steve, and we said, "Don't worry about Supabase, just keep building Postgrest and just making it awesome.

Um, and then obviously, if we have support requests or whatever, we can, we can ask him." Yeah. But, um, and the model just worked really well. And then we ended up, when Steve wanted to grow the Postgrest team, just hiring more people dedicated to work on that project. Well, you guys are doing so much for community and open source. One thing I do have in my mind is, like, what is the future of large commercial open source? Mm. Um, one, obviously there's a lot of slop, and, and- Yeah... uh, I'm sure you guys are dealing with that. But two, also your dependencies are now becoming, uh, optional. Like, you can just vibe code whatever you want. [laughs] Anything, yeah. Yeah, yeah.

So I, I, uh, whatever in your, is i-in your heads about the future of open source, I'd, I'd just love to hear it. I think the rewrite in Rust is a valid mechanism. I mean, someone came to me on our team, and they're like, "Oh, yeah, yeah, I'm rewriting everything in Rust that we're doing." Like, uh, and I was like, "To what level? Are you gonna rewrite the database? Are you gonna rewrite Postgres?" Postgres is not rewritten and, uh, is not in Rust, but I doubt down to that level you care. So there's, like, layers upon which you want to get to where things will be rewritten, and then you don't wanna be spending your actual innovation tokens or your actual tokens rewriting everything.

[laughs] So there's gonna be, like, a, a collapse of almost every tool probably into, let's say it's Rust, if that ends up being the, the ultimate way of writing things 'cause it's faster and more correct, and that will happen probably for most open source tools. But then after you've done this iteration, and everything's running on this highly effective language, then you're not gonna be rewriting it all of the things every single time. There'll be a few winners. A really classic one could be, say, Postgrest is written in Haskell, which is a very unusual language, and it's very hard to maintain. Could it be re-written in Rust now? Yeah, probably it could.

And the mechanism, the, the function is still valuable, and probably there'll be, if that happens, there'll be one Rust fork of Postgrest, and that will be the winner for that particular category. But you haven't done it yet. And your, and y- a lot of your stuff is Elixir as well. Yeah, so that's the other thing. Then you have to think on the infrastructure side. Elixir is, like, a really interesting language- Beam... because of the Beam. Yeah. So, so and then you start thinking when you're operating things, what is the right mechanism for operating, so. But now you can change paradigms eas- pretty much easily as long as you fit all your tests and, you know. Yeah, yeah. Well, that's a good one. I haven't thought a lot about would we switch Elixir to Rust.

It's just an option now. Yeah, Malta f- from Vercel is, uh, you know, one of his tweets that sticks in my mind. He- he's got lots of, uh, interesting thoughts. He set aside some amount of money and rewrite all of our dependencies, uh, at, at the C level or C++ level into Rust, um, because we just know it is better, it's safer, more correct- Yeah... what- whatever. And it is, it is a one-time upgrade of open source. Yeah. All the drivers, what- whatever you, you use. Um, so I, I thought that was, like, a pretty interesting call. I like that as an idea, yeah. Cool. I don't think it'll happen [laughs] because people love to do it themselves. Yeah. But, but as an idea, I, I would, like, we as a company would 100% get behind that. Awesome.

And then, uh, I think closing question, which I, I should have stuck in somewhere earlier, AI coding within Supabase, right? Uh, obviously you guys are big beneficiaries of coding agents outside. Yeah. Uh, what is the policy internally? How much are you spending? Is there, is there concern about spend? Anything of that sort. Supabase has always felt like a bunch of different startups, like, glued together.

Yeah. Yeah. Same salary regardless of where you live. If you're very good at what you do, please take a look at our job board. [laughs] Yeah. Can you, uh... Can you, like, one profile, just pick one, whatever is top of mind as, uh, as your highlight for what you're looking for and why? SREs. Okay. Yeah. Why? Because we're just growing at a ridiculous rate. We need all the SREs we can get. What is a quality that is... Like, you know, you, you probably get a lot of SRE applications and you don't look at them. Like, what, what stands out? The thing that we really care about is that you, you actually want to work remotely. Um, so it's- [laughs] This is a strange one, but yeah. [laughs] It, it's true.

We get a lot of quality people, but to fit in at Supabase, you kind of need to thrive in a remote-first environment, like a fully async environment. It's actually not for everyone. We learned that pretty early on. People would come in and they'd be like, "Oh, this is... Like, I need to see humans." My typical model of an SRE is, like, this gray beard that's, like, sitting in a dungeon somewhere and- Yeah, yeah... doesn't wanna talk to humans anyway. That's the perfect- Right. [laughs] That's the perfect person. [laughs] All of those people, please apply at Supabase. [laughs] Yeah. Great. Hopefully they listen to the podcast.

Yeah. Yeah, yeah. No, I mean- [laughs] Yeah, we, we, we probably do have a lot of, uh, gray beards, um, in SREs. I, I think one of... Uh, I have been on the record as saying AI SRE is one of the hardest categories for- Oh, yeah... for agent startups. I think I might be wrong soon, but so far it's been correct. Like- Okay, yeah... AI has not cracked SRE for, for whatever reason. No, no, no, no. Not yet. I agree. Yeah. Like, building software is, you know, debatable, solve problem. Maintaining, like keeping things running- Yeah... is a whole different beast. Well, okay. Well, thank you for keeping things running for everyone else. [laughs] You guys, congrats on your journey, man. Like, I'm so proud to be your friend- Thanks for this...

uh, and, and an investor and, and just, like, just see the progress. I cannot wait for the IPO, whenever that is. [laughs] Uh, it looks like it's gonna take some more time, but it'll, it'll be, it'll be fine. I, I want, really wanna see you guys, like, you know, do the whole ring the, ring the bell and, you know, building the next great open source company, right? Excellent. Definitely, man. Yeah. Congrats.