Supabase’s Plan for a Billion AI-Built Apps

WEBVTT

1 00:00:00.340 --> 00:00:04.820 Hello, it's me again. I hope, uh, you're enjoying these little episode intros that we're doing.

2 00:00:05.120 --> 00:00:14.000 Hopefully, it gives me a little chance to talk to you guys about why we're doing some of these episodes and what you should look out for in the interviews so you get more out of the interviews.

3 00:00:14.040 --> 00:00:26.320 We're also adding overlays that you can see pop up to keep it visually interesting, to explain things that maybe if you're newer to the industry you don't know, 'cause I don't have time to go over every single thing while we're having interviews.

4 00:00:27.100 --> 00:00:35.060 Um, so I was at Supabase Select recently, which is the Supabase company conference. Um, and there's a lot of stuff to cover.

5 00:00:35.320 --> 00:00:46.780 Uh, in particular, I think I should share that I have a very close relationship with this company. Um, I was an angel investor in their seed round six years ago, and they were my very, very first angel investment.

6 00:00:46.860 --> 00:00:57.340 They've done super well. They went from zero to 10 billion in re- in valuation in the last six years as a fully remote open source company, which is enormous.

7 00:00:57.520 --> 00:01:03.860 Uh, and they're pretty much becoming the standard for hosting Postgres. So I want to break down what, what makes them tick.

8 00:01:04.180 --> 00:01:09.940 Um, I also want to see why they did so well in my AEO ranking, uh, because it's not g- it's not a given for sure.

9 00:01:10.460 --> 00:01:21.880 Uh, and there's a lot of, uh, companies that are fighting for, uh, what it means to build databases for agents. They were also, I think, our very, very first sponsor for AI Engineer, the very first AI Engineer in 2023.

10 00:01:22.080 --> 00:01:36.900 Um, and they're now gonna be our first sponsors for Latenz Space, uh, which, because we're open for sponsorships and, uh, you know, who else to, to do this with, with companies that, and people that I'm familiar with and trust and can explain very well to, uh, to our audience.

11 00:01:37.320 --> 00:01:47.440 And so the main things I wanted to cover, they have, they have a whole recap here on their YouTube, which is very well executed and all that. Uh, but they have th- these three themes of build, operate, and scale.

12 00:01:47.880 --> 00:02:03.440 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.

13 00:02:04.040 --> 00:02:11.420 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.

14 00:02:11.740 --> 00:02:19.500 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.

15 00:02:20.120 --> 00:02:30.520 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.

16 00:02:30.980 --> 00:02:41.420 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...

17 00:02:41.740 --> 00:02:44.380 But declarative schemas is so important, right?

18 00:02:44.620 --> 00:02:54.040 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.

19 00:02:54.080 --> 00:03:02.840 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.

20 00:03:03.860 --> 00:03:14.760 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.

21 00:03:15.120 --> 00:03:27.280 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.

22 00:03:28.160 --> 00:03:35.900 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.

23 00:03:35.960 --> 00:03:44.180 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

24 00:03:45.220 --> 00:03:58.080 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.

25 00:03:58.500 --> 00:04:05.500 They've-- every year they ship, uh, some interesting pipeline work from like, you know, Stripe to, to, to, to other sort of integrations.

26 00:04:05.720 --> 00:04:15.480 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.

27 00:04:16.300 --> 00:04:23.380 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.

28 00:04:24.060 --> 00:04:33.000 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.

29 00:04:33.460 --> 00:04:38.040 Um, and really they're going after more and more scale obviously because they're grow- the customers are growing.

30 00:04:38.280 --> 00:04:49.480 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.

31 00:04:49.560 --> 00:05:00.700 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...

32 00:05:00.720 --> 00:05:06.820 So like there's a lot to get in there and honestly, uh, you can see my connection with the, with the founders.

33 00:05:07.360 --> 00:05:20.000 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.

34 00:05:20.500 --> 00:05:37.800 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.

35 00:05:37.920 --> 00:05:47.080 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.

36 00:05:47.500 --> 00:05:58.720 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.

37 00:05:59.020 --> 00:06:09.240 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?"

38 00:06:09.640 --> 00:06:21.060 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.

39 00:06:21.700 --> 00:06:35.620 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...

40 00:06:36.040 --> 00:06:44.440 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.

41 00:06:45.140 --> 00:06:53.080 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.

42 00:06:53.340 --> 00:07:02.220 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.

43 00:07:02.540 --> 00:07:13.120 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.

44 00:07:13.360 --> 00:07:26.380 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.

45 00:07:26.420 --> 00:07:34.680 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.

46 00:07:34.860 --> 00:07:38.420 So the first one, which we talked about last year, which was OrioleDB.

47 00:07:39.100 --> 00:07:46.740 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.

48 00:07:47.260 --> 00:07:56.040 The three innovations are basically, uh, performance, scaling up, and scaling out. Uh, Postgres touches on... Uh, sorry. OrioleDB touches on performance.

49 00:07:56.520 --> 00:08:09.140 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.

50 00:08:09.720 --> 00:08:16.660 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.

51 00:08:17.200 --> 00:08:25.880 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.

52 00:08:26.000 --> 00:08:37.840 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.

53 00:08:38.220 --> 00:08:48.560 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.

54 00:08:48.920 --> 00:09:00.740 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...

55 00:09:00.760 --> 00:09:10.740 the second one, so scaling up. I see. So that's, that's Mu- that's Multigress. Yes. So Multigress is the second big announcement. Basically, we looked at the Postgres ecosystem, and we looked at all the operators.

56 00:09:10.780 --> 00:09:20.960 An operator, for those who aren't familiar, when you run Postgres, you start with a single node, but then with almost any database, you wanna run multiple nodes, and you want things like failover.

57 00:09:21.020 --> 00:09:26.660 You want it to manage backups. So usually you have this thing called an operator which can run the entire cluster for you.

58 00:09:27.120 --> 00:09:35.200 So there's quite a few operators in the Postgres ecosystem, and what we had found was that under, um, certain conditions, every single one of them could lose data.

59 00:09:35.580 --> 00:09:42.280 And so we decided with Multigress to do a ground up rewrite of the entire structure of a, of a operator.

60 00:09:42.700 --> 00:09:53.300 And so Multigress is essentially, while our initial goal was to become or give sharding to the Postgres community, what has turned out to be is this entire operating system.

61 00:09:53.720 --> 00:10:02.800 And we actually ran it, uh, again through very rigorous testing, and we've done a thou- uh, 1,000 sequential failovers, and it didn't lose any data.

62 00:10:02.860 --> 00:10:13.740 And we've tested it in basically every way we can, clobbering it to see whether it will ever lose data. So we're very confident now in its stability, and so we're releasing that one as well tomorrow.

63 00:10:13.779 --> 00:10:24.180 It's completely open source. It will be on our platform for customers and, uh, invite only, so people who are running serious workloads can come to us, and we'll start helping them to scale up.

64 00:10:24.580 --> 00:10:33.680 Is this the full completion of Multigress being fully sort of parallel to Vitess, or is there still le- something left to go? No, actually, yeah, it was quite interesting.

65 00:10:33.740 --> 00:10:42.380 So Suku, um, originally created, uh, Vitess in, for the MySQL ecosystem, but specifically for sharding.

66 00:10:43.040 --> 00:10:57.380 And actually, when he came along, he said, "Well, the number one goal that we wanna do for this time," because we're always catching up on compatibility when we use Vitess, you always had to change things as you move to sharded, so change some queries.

67 00:10:57.760 --> 00:11:01.340 So he said, "I wanna focus only on compatibility first."

68 00:11:01.780 --> 00:11:12.980 And so the way we've designed it is in such a way, this ground up rewrite, is specifically so that we know when we release sharding, you actually won't have to change anything for your database.

69 00:11:13.160 --> 00:11:23.020 You'll continue to run the same queries, and Multigress will just be able to handle it. Zero changes to your code. I mean, that's a incredible vision, and the, the fact that you've been able to do it in two years.

70 00:11:23.700 --> 00:11:26.220 That's one year. One year. Um, so we haven't released sharding yet.

71 00:11:26.840 --> 00:11:38.520 I've seen the early prototypes and the designs for how we're doing it, and it should, should, some of it should land by probably early next year, depending on some of the things that are happening on our platform.

72 00:11:38.800 --> 00:11:46.240 We've got a lot going on. But, um, but we're very confident. We're, and we're, we're happy with the progress. It's, it's been pretty good. Yeah.

73 00:11:46.560 --> 00:11:56.480 I mean, I think this is the last piece for the total domination of Postgres. Yeah. Right? Yeah, that's, uh, that's it. And for those who don't know, you know, look up Vitess, uh, it powers YouTube, powers, what else?

74 00:11:56.520 --> 00:12:08.116 GitHub, uh, everything. Yeah. If workloads were ever to shift away from MySQL, you need this. To, to Postgres Yeah And that's then for Supabase to win. And hopefully it, it's not just for Supabase customers, right?

75 00:12:08.236 --> 00:12:15.376 Like, it's fully open source It's community things It's anyone who's- Who's involved... who's running Postgres- Yeah... at scale, um, should be able to use Multigraph.

76 00:12:15.596 --> 00:12:21.516 Yeah, I'm sure, like, every coalition, every company that's involved in Postgres in some way is contributing to this thing, right? 'Cause we all want this. Yeah, yeah.

77 00:12:21.596 --> 00:12:32.376 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.

78 00:12:32.416 --> 00:12:43.376 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.

79 00:12:44.256 --> 00:12:54.956 At the moment, we launch about five million databases on our platform every single month, and kinda the future where we know we're going, we think that we're going to have to launch billions per month.

80 00:12:55.296 --> 00:12:58.556 And so scaling out is a whole different beast.

81 00:12:59.116 --> 00:13:12.596 So we decided that we would have to look at some new architectures and people within the space, the agentic database space, and there's no other leading database provider who does agentic well than Turso.

82 00:13:13.396 --> 00:13:24.716 And so actually tomorrow, and when this is announced, it would've come out, where we have announced that we've acquired Turso to help us power agentic databases. Yeah, that's an incredible...

83 00:13:24.876 --> 00:13:33.196 I- it's also, like, they're, they're primarily SQLite based, but it doesn't really matter, right? Like, it's, it's more about the architecture of, like, I guess...

84 00:13:33.776 --> 00:13:40.616 I don't have the word to describe how basically, like, broadcasts, is that the word? For, for scale out, right?

85 00:13:40.656 --> 00:13:53.036 Like, um, there's the idea for them scaling SQLite was that they would have, like, a single writer and, like, multiple readers. Yeah, so, so SQLite itself, it's only a single writer. It's a file format.

86 00:13:53.136 --> 00:13:58.196 Um, so generally it's not thought of as a- Like sort of a MVCC for SQLite. Yeah, ex- exactly, exactly.

87 00:13:58.476 --> 00:14:09.536 So then they built this Turso architecture, which basically builds MVCC for SQLite so that you can have multiple writers. Yes. And then it's also- And because SQLite is, has a weird license, they had to fork it?

88 00:14:09.596 --> 00:14:16.296 They had to fork it. [laughs] They originally started with Libsql, and then they- [laughs]... they came along, and they did this other really cool architecture.

89 00:14:16.736 --> 00:14:27.656 And actually now the architecture, actually it was very commonly discussed a few weeks ago, well, on S3. So some of the innovations in the database space are actually coming out of what's happening in object storage.

90 00:14:28.016 --> 00:14:34.556 So object storage has added a bunch of new features, which basically mean that you can start using it a little bit like a, a database.

91 00:14:34.876 --> 00:14:43.136 And because object storage itself is, I think it's 11 nines of, of data- Durability... durability. Yes Then- Which is not reliability, but whatever... yeah, not reliability.

92 00:14:43.556 --> 00:14:51.156 But it, it gives you what you need for, for, um, basically what you would typically have for a cluster of databases.

93 00:14:51.536 --> 00:15:02.136 It's kind of like your cluster underneath, so you can start doing some cool things like, yeah, there's quite a few popular databases now, or data use cases that are building directly to- on top of object storage.

94 00:15:02.596 --> 00:15:05.776 It's got some limitations, um, in our worldview.

95 00:15:06.016 --> 00:15:20.476 It's probably good for these smaller use cases in terms of transactions, because when you've got a transactional database, you've gotta do a lot of reads and writes, and so in this case, you probably don't wanna be doing something at massive scale on this, and that's why we have Multigress.

96 00:15:20.896 --> 00:15:33.236 But for these, what we're seeing, where we need to get to, these billions of databases, what we need them to be is a file format. And so that's where SQLite was super interesting. King. It's dominating.

97 00:15:33.356 --> 00:15:47.976 And actually, the thing that really piqued our interest, and I reached out to Glover, h- I said, uh, he released a blog post on how they've done their rewrite, basically makes the-- we have a pathway to get Postgres compatibility with their file format.

98 00:15:48.356 --> 00:15:52.156 So we're ac- acquiring them, and we're actually gonna keep the Turso brand.

99 00:15:52.236 --> 00:16:03.256 So Turso's gonna focus on SQLite, and then we're going to introduce this new, smaller instance type on Supabase called Supabase Spark, and it'll be compatible with all our libraries.

100 00:16:03.616 --> 00:16:13.436 And you can have one database for any idea, any prototype that, that you want, and when it takes off, then you just upgrade to full Postgres databases, all the way up to Multigress.

101 00:16:14.176 --> 00:16:23.796 I know you've been working on this for a while. This is the dream, right? Yeah, yeah. There's been-- Both scaling up and scaling down has been one of the things that we've been trying to figure out how to do. Yeah.

102 00:16:24.116 --> 00:16:31.056 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.

103 00:16:31.196 --> 00:16:40.596 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...

104 00:16:41.216 --> 00:16:45.856 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?

105 00:16:45.996 --> 00:16:51.216 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.

106 00:16:51.876 --> 00:17:01.096 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.

107 00:17:01.156 --> 00:17:12.256 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?

108 00:17:12.316 --> 00:17:23.976 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.

109 00:17:24.496 --> 00:17:36.796 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...

110 00:17:36.836 --> 00:17:46.776 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.

111 00:17:47.276 --> 00:17:55.236 And I think, um, there's a much larger market of almost, like, consumer, like, builders, right? So people- Mm...

112 00:17:55.276 --> 00:18:02.276 inside companies, people who wanna launch their own startups, which is gonna mean a lot more databases, which- [laughs]

113 00:18:02.316 --> 00:18:14.632 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.

114 00:18:15.172 --> 00:18:25.932 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.

115 00:18:26.172 --> 00:18:34.312 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.

116 00:18:34.532 --> 00:18:43.552 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.

117 00:18:44.032 --> 00:18:50.432 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?

118 00:18:50.472 --> 00:18:57.972 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...

119 00:18:58.012 --> 00:19:04.212 We've got a small- Okay [laughs]... team of, uh, teenagers now who help me out. [laughs] You do workshops, right?

120 00:19:04.372 --> 00:19:19.632 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?

121 00:19:20.012 --> 00:19:30.152 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.

122 00:19:30.152 --> 00:19:38.992 [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.

123 00:19:39.132 --> 00:19:43.572 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.

124 00:19:43.572 --> 00:19:55.592 [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.

125 00:19:55.612 --> 00:20:05.932 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.

126 00:20:05.972 --> 00:20:17.412 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.

127 00:20:17.491 --> 00:20:26.012 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.

128 00:20:26.332 --> 00:20:31.652 So then when we onboarded new marketing people, we were like, "Here's the strategy." [laughs] That was written by someone else. Yeah.

129 00:20:32.212 --> 00:20:39.372 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.

130 00:20:39.432 --> 00:20:47.432 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.

131 00:20:47.452 --> 00:20:56.732 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.

132 00:20:57.132 --> 00:21:09.012 [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.

133 00:21:09.072 --> 00:21:18.972 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...

134 00:21:20.492 --> 00:21:31.852 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.

135 00:21:32.292 --> 00:21:43.172 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.

136 00:21:43.232 --> 00:21:54.672 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...

137 00:21:55.052 --> 00:21:58.892 Dang has super high credibility. Yeah No one, no one really, like, you know, uh, questions him.

138 00:21:59.292 --> 00:22:12.152 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.

139 00:22:12.432 --> 00:22:17.472 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?

140 00:22:17.472 --> 00:22:27.572 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.

141 00:22:27.612 --> 00:22:32.772 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.

142 00:22:32.812 --> 00:22:40.592 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...

143 00:22:40.652 --> 00:22:50.312 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...

144 00:22:50.352 --> 00:22:59.092 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.

145 00:22:59.112 --> 00:23:08.292 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.

146 00:23:08.332 --> 00:23:17.612 We weren't ready to be launched on Hacker News. And you didn't consider yourself a Firebase alternative. Oh, yeah. So that was the day that we changed the tagline. We had the tagline- What was the original tagline?

147 00:23:17.732 --> 00:23:19.032 It was real-time Postgres- Ah...

148 00:23:19.072 --> 00:23:33.512 because what had happened was I wanted, like, we, we were like, "Oh, let's build a Postgres company," but in my previous company, I had to migrate from Firebase to Postgres, but I needed a real-time f- feature, and so I ended up building this real-time engine on top of Postgres.

149 00:23:33.792 --> 00:23:41.752 And so we thought, "Oh, that's the killer thing. Like, it's getting some open source traction. This real-time engine is kinda cool." So we were like, "Oh, real-time Postgres." No one cared.

150 00:23:42.172 --> 00:23:49.732 No one cared when we had a website, and so, like, we were like, "Let's change it to open source Firebase alternative," knowing that that was the reason why I switched.

151 00:23:50.232 --> 00:23:56.852 And it was, I think, only 12 hours later or something like that- Yeah, yeah... that this article went up. Yeah Yeah. Still the top of all time.

152 00:23:57.212 --> 00:24:13.156 And at the time, we were s- running on top of DigitalOcean, and every new database was two droplets. So two $5 a month droplets. [laughs] Why two? Um, one for the, uh, database and one for the APIs. [laughs] Right.

153 00:24:13.376 --> 00:24:19.756 Because we- 'Cause we wanted to upgrade the APIs and decouple- Independently, yeah. And we only had a handful of, like, test users at that point. Yeah, yeah.

154 00:24:19.776 --> 00:24:28.236 And then all of a sudden we were- I think we only had eight users. Yeah. We went from eight to eight hundred on that, that launch alone. Yeah. And then we burned through...

155 00:24:28.276 --> 00:24:34.436 I think we had, like, 10K of Digital, uh, Ocean credits, which we just burned through in basically 48 hours.

156 00:24:35.036 --> 00:24:43.856 Um, and then we'd just gone into YC, so AWS was like, "Oh, we'll give you," you know, whatever it was, "150K in credits if you migrate to us."

157 00:24:44.236 --> 00:24:53.656 And so we spent the first week of being in YC migrating 800 databases to a different cloud, which, you know, if you've ever migrated one database, you know how painful it is.

158 00:24:54.116 --> 00:24:58.856 It's like migrating 800 is not how you wanna spend your first [laughs] week in YC. That's crazy.

159 00:24:58.896 --> 00:25:10.896 Do you remember that we could have been on Digital-- Like, bas- basically, we should still be building on DigitalOcean, 'cause when you get into YC, they have this credits program, like, huge number of credits f- on DigitalOcean at the time.

160 00:25:11.456 --> 00:25:18.536 And we're like, "Hey, can we have these credits? Because we're, like, we're blowing up here." And they said, "Oh, no, you've already used the $10,000 credit program.

161 00:25:18.956 --> 00:25:28.276 We can't give you this other one for three hundred thousand." So we're like, "Well, we have to migrate to AWS." [laughs] And so the biggest lost opportunity of DigitalOcean ever- Being too strict, yeah...

162 00:25:28.396 --> 00:25:35.176 didn't-- I know our bills with AWS now, this is, like- Nothing... probably one of their biggest customers ever that they lost- Yeah... from being too strict about that.

163 00:25:35.176 --> 00:25:45.756 But because it was COVID, it was a time when everyone was worried about, like, what's- Yeah... gonna happen. True. So I think they were like, "We don't wanna overextend ourselves." Um- It's fair. Yeah. Yeah.

164 00:25:45.916 --> 00:25:53.936 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.

165 00:25:54.236 --> 00:26:04.316 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.

166 00:26:04.356 --> 00:26:12.936 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.

167 00:26:12.996 --> 00:26:23.596 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.

168 00:26:24.116 --> 00:26:32.136 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.

169 00:26:32.136 --> 00:26:39.196 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.

170 00:26:39.356 --> 00:26:50.716 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.

171 00:26:50.776 --> 00:26:58.956 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.

172 00:26:59.116 --> 00:27:04.016 We work with them already on Gemini, and we wanna make sure that we're embedded deeply into all of their products.

173 00:27:04.356 --> 00:27:14.576 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.

174 00:27:15.096 --> 00:27:24.096 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,

175 00:27:25.076 --> 00:27:33.816 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.

176 00:27:34.016 --> 00:27:42.136 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?

177 00:27:42.156 --> 00:27:47.636 [laughs] Yeah, yeah. And then S3 Express as well, that was a big game changer, um, because- Why?

178 00:27:48.056 --> 00:27:58.476 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...

179 00:27:58.756 --> 00:28:09.336 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.

180 00:28:09.396 --> 00:28:17.796 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.

181 00:28:17.836 --> 00:28:27.735 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?

182 00:28:28.576 --> 00:28:38.136 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...

183 00:28:38.176 --> 00:28:47.976 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.

184 00:28:48.296 --> 00:28:56.496 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.

185 00:28:56.916 --> 00:29:05.756 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?

186 00:29:05.976 --> 00:29:15.156 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.

187 00:29:15.236 --> 00:29:21.176 At, at the start it was like you choose your database, you build your product. At some point, you, um, add in billing.

188 00:29:21.616 --> 00:29:32.456 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.

189 00:29:32.496 --> 00:29:42.516 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.

190 00:29:42.916 --> 00:29:52.036 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.

191 00:29:52.056 --> 00:29:54.396 People are making real things that generate real money.

192 00:29:54.876 --> 00:30:15.200 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.

193 00:30:15.600 --> 00:30:28.600 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.

194 00:30:28.800 --> 00:30:36.620 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.

195 00:30:37.060 --> 00:30:47.340 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.

196 00:30:47.380 --> 00:30:57.960 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.

197 00:30:58.240 --> 00:31:05.540 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.

198 00:31:05.560 --> 00:31:12.540 [laughs] We did this talk. Uh, basically, all these embedding databases were coming out, vector databases, create the vector.

199 00:31:12.700 --> 00:31:22.620 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.

200 00:31:23.080 --> 00:31:29.900 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.

201 00:31:30.000 --> 00:31:40.720 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.

202 00:31:40.960 --> 00:31:49.480 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.

203 00:31:49.880 --> 00:31:58.980 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?

204 00:31:59.420 --> 00:32:08.040 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.

205 00:32:08.640 --> 00:32:17.600 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.

206 00:32:17.800 --> 00:32:24.020 They were doing toy things, but still, it fit our mission. We just had to believe that over time, these tools would become better.

207 00:32:24.580 --> 00:32:37.700 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.

208 00:32:38.060 --> 00:32:47.200 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.

209 00:32:47.200 --> 00:32:51.320 [laughs] So, um, so- Here, I got, I got one of your charts. There we go. Yeah.

210 00:32:51.360 --> 00:32:57.600 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...

211 00:32:57.600 --> 00:33:07.520 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.

212 00:33:07.660 --> 00:33:16.120 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.

213 00:33:16.220 --> 00:33:26.560 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.

214 00:33:26.620 --> 00:33:37.920 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.

215 00:33:38.040 --> 00:33:47.100 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.

216 00:33:47.140 --> 00:33:56.420 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.

217 00:33:56.880 --> 00:34:04.500 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?

218 00:34:04.600 --> 00:34:12.260 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.

219 00:34:12.800 --> 00:34:18.600 And I think it is interesting, like you had to be in the market for three to four years, exactly your- Yeah...

220 00:34:18.660 --> 00:34:30.360 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.

221 00:34:30.360 --> 00:34:34.160 [laughs] I mean, we couldn't have timed it better. Like it was a complete accident.

222 00:34:34.540 --> 00:34:43.160 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...

223 00:34:43.420 --> 00:34:54.320 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.

224 00:34:54.360 --> 00:35:03.440 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?

225 00:35:03.660 --> 00:35:17.620 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.

226 00:35:17.660 --> 00:35:25.260 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.

227 00:35:25.520 --> 00:35:33.300 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.

228 00:35:33.360 --> 00:35:38.400 So, so last year at Supabase Select, we announced that we were gonna do a crowdfunding round.

229 00:35:38.440 --> 00:35:44.400 It's something we've always wanted to do, and we should have done it much earlier on, but we just, you know, had so many other things going on.

230 00:35:45.080 --> 00:35:57.820 And basically then we spent the last year fighting with the lawyers who are saying, you know, for the number of people we wanna be involved, there is no mechanism to do it without people being accredited investors.

231 00:35:58.320 --> 00:36:07.180 Until we asked them the question, "Well, what if they don't invest? What if we just give them the stock?" [laughs] And they were like, "Yeah, we'll look into it."

232 00:36:07.220 --> 00:36:12.800 And then they came back, and they were like, "Actually, there is a mechanism [chuckles] where if you really just wanna give the stock away, you can."

233 00:36:13.300 --> 00:36:20.480 And so now we've earmarked, like, um, a million dollars in Supabase stock that we're basically gonna give to a bunch of the community members.

234 00:36:20.820 --> 00:36:29.300 The first community member we're, we're gonna stand up on stage and sort of announce is, is this guy, his name's Gary Austin in the Supabase Discord.

235 00:36:30.320 --> 00:36:43.560 And this guy's, like, a legend of the Supabase community Yeah He-- We looked into his stats, and Gary has answered, uh, 20,000, in the last six years, 20,000, uh, questions in the Supabase Discord.

236 00:36:44.220 --> 00:36:57.620 And the crazy thing is 13,000 of them he answered within the first hour of them being asked. Hmm. And he's not a Groq bot. Like- [laughs]... he's a real human. [laughs] Yeah. Have you met him? What, what does- Yeah.

237 00:36:57.640 --> 00:37:06.900 We- What does he do? He's real. We, we have met him. He, he was basically, like, retired- Yeah... and just interested in Supabase. Gary Austin. Gary Austin. Or Gary Abbott? Yeah, Gary Abbott- Yeah... is his real name.

238 00:37:06.940 --> 00:37:16.620 Okay, this guy. Yeah. [laughs] Yeah. Okay. And, and he loved Supabase because he basically built his own som- sommelier, um, app 'cause he loves red wine. Mm-hmm.

239 00:37:16.960 --> 00:37:22.020 Um, and then was just getting a kick out of, you know, he's got the Supabase logo on his Christmas tree.

240 00:37:22.040 --> 00:37:30.440 [laughs] And he's also, like- By the way, this is my favorite Supabase memes, like, just, like, you reminding me that, like- [laughs] Yeah... I'm about to run out of my Supabase.

241 00:37:30.840 --> 00:37:41.660 That's the announcement is we're basically, like, um, anyone who's a, a legend in the Supabase ecosystem we can now- Yeah... um- Is Discord still a strong... I mean, I, I got mixed feelings about Discord.

242 00:37:41.720 --> 00:37:51.900 It, it is, it is the best chat. You know, Slack sucks. But also it is not public, right? So all the work in there is- Yeah... not scalable. Yeah. Have you found anything to solve that?

243 00:37:52.140 --> 00:38:03.400 There was some tools that we looked at, uh, in the early days. We, we decided not to do it because you still needed to police, like, what things you wanted to be public and, and what things you didn't.

244 00:38:03.820 --> 00:38:13.040 But the original debate that I remember we debated a lot was, okay, we wanna host the community. Do we do a Discord? Do we do a Slack community? Do we do a Reddit?

245 00:38:13.580 --> 00:38:24.060 And we tried a few different options, but we landed on this thing of, like, Discord users are gonna use Discord, Reddit users are gonna use Reddit, and we basically had to do all of the above. Just be where they are.

246 00:38:24.140 --> 00:38:31.920 Yeah. And be where they are, yeah. That's the answer. Which there was a lot of work. [laughs] Yeah. And it worked out, right? I mean- But-... that was the right answer...

247 00:38:31.940 --> 00:38:42.220 but then there's a lot of, like, great tools that started popping up, right? Um, Orb- I don't think anyone has la- has, uh, s- stood the test of time, let's call it. No. Um, but yeah. There, there- But, but it helped...

248 00:38:42.220 --> 00:38:52.140 there are, there are tools that are Discord mirrored to public Q&A site. That is a thing that people do. Mm-hmm. I don't know if it's been successful for anyone. Yeah. Yeah. And, and, like, Common Room, um, helped a lot.

249 00:38:52.140 --> 00:39:05.820 Actually, the only thing that we did was for ourselves. So basically anyone who had asked a legitimate question on any one of these, we then would basically scrape the question and the response into our own website.

250 00:39:06.060 --> 00:39:16.480 No, actually, we scraped it kind of into GitHub Discussions- Mm-hmm... and mirrored that onto our website. And so the content that was getting slurped up- Ah, used GitHub as a CMS. Yeah, we used it- Yeah... as a CMS.

251 00:39:16.540 --> 00:39:26.140 Exactly. Yeah. It was a great strategy because everyone-- Ultimately, everyone was asking the same questions over and over again. So we were just- It is true... like, "Okay. Well, we'll surface it each time." Yeah.

252 00:39:26.460 --> 00:39:32.240 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.

253 00:39:32.820 --> 00:39:41.800 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.

254 00:39:42.160 --> 00:39:53.840 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...

255 00:39:54.740 --> 00:39:58.820 we don't put any tracking in our open source offering. You can use it.

256 00:39:58.980 --> 00:40:10.080 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?"

257 00:40:10.140 --> 00:40:21.100 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?

258 00:40:21.100 --> 00:40:30.180 '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.

259 00:40:30.600 --> 00:40:39.060 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.

260 00:40:39.560 --> 00:40:51.040 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.

261 00:40:51.320 --> 00:41:03.020 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.

262 00:41:03.220 --> 00:41:13.040 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.

263 00:41:13.060 --> 00:41:21.060 [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.

264 00:41:21.100 --> 00:41:27.920 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?

265 00:41:27.960 --> 00:41:38.480 Like, what, what are-- What is that story apart from, like, give me a dashboard? Well, first of all, we should talk about sales strategies. Okay. So Supabase is a very famously P&G company.

266 00:41:38.580 --> 00:41:52.040 [laughs] Like, we give a shit about... We have 60% of every YC batch use Supabase. We focus almost all of our attention on this, like, becoming an index company. Basically, our role model here is Stripe.

267 00:41:52.180 --> 00:42:01.720 We wanna follow in their footsteps. They're one of the only developer tools that we've ever seen that has withstood more than a decade of moving our market and still keeping that developer love.

268 00:42:01.820 --> 00:42:11.340 So that's the first thing. So to do that, all we started doing when we did sales was layer on what we call product-led sales. So we look instead for, like, product triggers.

269 00:42:11.380 --> 00:42:22.400 It could be, like, a, a logo, someone who we think is really important, but more importantly, they could be scaling up. They could be hitting some limits, having some errors, paying a certain amount.

270 00:42:22.480 --> 00:42:32.300 We've got a bunch of these triggers, and then AI came along, and we're like, "Oh, we can do some interesting stuff from this." So let's say we have 1,000 product triggers every single week.

271 00:42:33.232 --> 00:42:45.752 We send off a thousand emails to people who are having problems, and say five hundred of them respond. We take the cohort who don't respond, and we use them as a control group. Okay. And we measure how much they grow.

272 00:42:46.252 --> 00:42:56.132 Turns out they grow twenty-five percent m-month over month. Then we take all those who do respond, and we say to our sales team, "The only thing you need to do is get on a call with these people and help them."

273 00:42:56.612 --> 00:43:02.512 And because we've got all their traffic patterns, we can give them a script, we know what their problem is. AI's been great on this.

274 00:43:03.212 --> 00:43:11.272 And, uh, we give them a script just to help out, and it turns out just getting on a call and helping people out, their account then grows a hundred and twenty-five percent.

275 00:43:11.652 --> 00:43:16.432 So that a hundred percent uplift, the incremental uplift, is what we compensate our salespeople.

276 00:43:16.812 --> 00:43:22.592 Traditionally, you would say, "Get on a call with th-these people and just sell them shit," and s- then you get the full quota.

277 00:43:22.832 --> 00:43:34.612 But because we know our database grows, our platform grows at twenty-five percent month over month on this cohort of users, then we know that our sales force are adding this one hundred percent incremental uplift.

278 00:43:35.112 --> 00:43:43.012 So that's how we do our sales function. It's always focused on these logos or very specific, uh, product needs.

279 00:43:43.372 --> 00:43:54.852 So yeah, the likes of the enterprise is usually coming in because they're wanting to launch lots and lots of databases, or almost never it'll be a transfer of their production database.

280 00:43:55.312 --> 00:44:08.612 So migrating databases from, say, MySQL to Postgres or from, say, one scalable database to another scalable database almost never happens. And that was the core thesis of Supabase at the start.

281 00:44:09.092 --> 00:44:20.712 We knew that every single Postgres company we had looked at or database company we had looked at in the past that had tried to do migrations had failed over time. You just can't convince people to migrate.

282 00:44:20.872 --> 00:44:29.852 Even if you build a bigger, better, more scalable Postgres, what you have to do is land with them on day one and then make sure that they stick around to day one thousand, day ten thousand.

283 00:44:30.252 --> 00:44:42.572 So all of our focus is winning these enterprise logos on, on day one. Greenfield. Greenfield, yeah. They're still not migrating? I guarantee they'll almost never migrate. Wow. There's a cohort that will need Multigress.

284 00:44:42.892 --> 00:44:55.992 We talked to some logos who are looking at using it open source. They're very interested on making sure that, um, they can basically help us direct because they're running into problems of Postgres at scale.

285 00:44:56.532 --> 00:45:08.352 But for the next, say, five years, almost all of our focus on the enterprise side is making sure that we can handle greenfield use cases, which now in the enterprise it turns out to be a very- Hmm...

286 00:45:08.352 --> 00:45:20.452 prolific use case because they want all of their product people to be developing, innovating as much as they can. So again, it's kind of perfect timing for us because that's what we've been focused on for so long.

287 00:45:20.552 --> 00:45:30.232 This control tower is, like, one of our biggest things. Multigress of course gives us the, the opportunity to win these large workloads if and when they want to ever migrate.

288 00:45:30.572 --> 00:45:40.672 But it won't ever be one of our core focuses, or at least not for the next, say, five years. We're gonna have to focus really heavily on just landing inside these enterprises and then expanding up.

289 00:45:41.032 --> 00:45:50.652 The other thing that's changed is we obviously put in the past all of our focus into developers. So it was like selling to the developers, enable them as a champion, they help you sell internally.

290 00:45:51.392 --> 00:46:08.072 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.

291 00:46:08.332 --> 00:46:11.352 Oh, that's great. Yeah. Mentioning Lovable, mentioning Bolt.

292 00:46:12.032 --> 00:46:21.632 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.

293 00:46:22.152 --> 00:46:32.352 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?

294 00:46:32.352 --> 00:46:41.912 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...

295 00:46:42.032 --> 00:46:46.932 front-end tool. Yeah. Um- Yeah... not that I don't think Lovable is sticky. I think it's amazing.

296 00:46:46.972 --> 00:46:55.492 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.

297 00:46:55.872 --> 00:47:06.972 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.

298 00:47:07.332 --> 00:47:17.392 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.

299 00:47:17.432 --> 00:47:26.032 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.

300 00:47:26.072 --> 00:47:38.692 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...

301 00:47:38.752 --> 00:47:45.392 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...

302 00:47:45.392 --> 00:47:55.612 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.

303 00:47:56.072 --> 00:48:09.432 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...

304 00:48:09.452 --> 00:48:14.252 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.

305 00:48:14.392 --> 00:48:19.832 I heard some crazy stat like that it might be different now 'cause they're trying to, Oracle are trying to move everything to the cloud.

306 00:48:19.872 --> 00:48:35.080 But I think it was like one year ago, someone said that only five percent of Oracle workloads by revenue ran in the cloud. And so, like almost everyone was saying, "Oh, let's rip out Oracle. Let's do migrations."

307 00:48:35.560 --> 00:48:47.040 People... That's what I mean, though. Yeah. Like, when I look through the history of database providers who have tried to run this strategy, it sounds amazing, and it's something that we'll of course try to do ourselves.

308 00:48:47.300 --> 00:48:57.800 Why would we not try to do this? But it is incredibly difficult, and I think you would just wanna focus, first of all, on landing inside the enterprise. Uh, yeah, that makes sense. Yeah.

309 00:48:57.840 --> 00:49:00.700 It's, it's one of those things where, like, you know, when, when I s- see...

310 00:49:00.740 --> 00:49:09.340 When I talk to coding agent people all day long, the great promise is that you can finally migrate off of all this legacy software that you have. Yeah. Right? And, like, hopefully- That's true...

311 00:49:09.360 --> 00:49:18.720 that, that means that you cut over everything. That's true, because one of the things that actually makes Oracle so sticky is, like, you end up writing these million-line stored procedures.

312 00:49:18.760 --> 00:49:29.020 Like, you stuff everything into your database. And so actually probably rewriting that is, these days, a lot easier. Right. Right. It's... So, like, hopefully coding agents are the boom. Uh, maybe they're not there yet.

313 00:49:29.120 --> 00:49:31.280 Maybe you guys have to become more on-prem.

314 00:49:31.700 --> 00:49:42.160 Most of the problem around migrations is actually not, not rewriting it, and you can do migrations, for example, on the data analytics side quite easily from one, from one warehouse to another.

315 00:49:42.580 --> 00:49:54.420 The problem is that usually your OL- OLTP database is powering your most revenue-generating application. So a minute of downtime means that you end up with basically dollars lost.

316 00:49:54.740 --> 00:50:08.920 And so then what you end up doing is having to run these two things in parallel, and then the risk of this is that you get this split brain where you've got some data running on one side, some on another, and if you try to reconcile s- split brain, anyone who's tried to do it, it's a very hard problem.

317 00:50:09.000 --> 00:50:18.640 So the risks are just so much higher when it comes to OLTP that a lot of the time people just start doing... splitting the workloads off, and they'll start new workloads.

318 00:50:19.020 --> 00:50:28.380 They'll do a lot of operations here, and hopefully they'll migrate things over. Yeah. Usually then you've got two problems, though. [laughs] Well, speaking of OLTP, I mean, I, I...

319 00:50:28.460 --> 00:50:41.100 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.

320 00:50:41.320 --> 00:50:48.560 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.

321 00:50:48.660 --> 00:50:56.620 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?

322 00:50:56.660 --> 00:51:05.760 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.

323 00:51:05.780 --> 00:51:12.740 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...

324 00:51:13.080 --> 00:51:22.820 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.

325 00:51:23.240 --> 00:51:35.580 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.

326 00:51:35.640 --> 00:51:40.820 So they're a foundation- [indistinct]... that are developing. Yeah. So they're building a committee, um, around the open source.

327 00:51:41.140 --> 00:51:54.120 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.

328 00:51:54.140 --> 00:52:00.400 It's like a professor in Netherlands? Yeah, exactly. They live, they live and work out of the Netherlands, which is pretty cool.

329 00:52:00.940 --> 00:52:12.460 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...

330 00:52:13.060 --> 00:52:24.040 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.

331 00:52:24.320 --> 00:52:36.260 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.

332 00:52:36.340 --> 00:52:46.740 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.

333 00:52:46.900 --> 00:52:56.480 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...

334 00:52:56.600 --> 00:53:05.140 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.

335 00:53:05.540 --> 00:53:17.800 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...

336 00:53:17.800 --> 00:53:27.360 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.

337 00:53:27.380 --> 00:53:36.680 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.

338 00:53:36.680 --> 00:53:46.640 [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...

339 00:53:46.700 --> 00:53:54.620 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.

340 00:53:54.660 --> 00:54:02.520 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?

341 00:54:03.100 --> 00:54:13.800 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

342 00:54:14.720 --> 00:54:21.840 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.

343 00:54:22.120 --> 00:54:27.640 Now we've got the primitive that would unlock the ability to do long-running durable workflows.

344 00:54:28.300 --> 00:54:38.580 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?

345 00:54:38.620 --> 00:54:49.624 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.

346 00:54:49.644 --> 00:54:51.484 [laughs] Pretty much. And be prolific.

347 00:54:51.544 --> 00:55:03.104 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.

348 00:55:03.584 --> 00:55:14.364 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.

349 00:55:14.424 --> 00:55:21.224 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.

350 00:55:21.264 --> 00:55:29.244 [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?

351 00:55:29.284 --> 00:55:36.584 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.

352 00:55:37.044 --> 00:55:47.744 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.

353 00:55:48.124 --> 00:55:58.044 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.

354 00:55:58.224 --> 00:56:05.664 [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.

355 00:56:05.964 --> 00:56:14.524 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.

356 00:56:14.604 --> 00:56:25.504 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?

357 00:56:26.084 --> 00:56:38.004 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.

358 00:56:38.044 --> 00:56:44.704 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.

359 00:56:45.164 --> 00:56:57.744 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.

360 00:56:57.784 --> 00:57:01.724 Like, is it a billion? Is it two billion? Is it... You know, you could just-- we could just keep going.

361 00:57:02.144 --> 00:57:09.644 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]...

362 00:57:09.704 --> 00:57:20.264 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...

363 00:57:20.304 --> 00:57:29.364 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.

364 00:57:29.704 --> 00:57:39.464 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?

365 00:57:39.564 --> 00:57:43.324 And then just this whole market just exploded, so let's see what happens.

366 00:57:43.624 --> 00:57:52.484 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.

367 00:57:52.524 --> 00:57:58.404 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.

368 00:57:58.444 --> 00:58:07.824 It's just, like, a ridiculous growth curve from the start of this year. Amazing. Um, what are the open questions that you have?

369 00:58:07.904 --> 00:58:19.784 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.

370 00:58:20.104 --> 00:58:28.464 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...

371 00:58:28.684 --> 00:58:35.284 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."

372 00:58:35.844 --> 00:58:47.564 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.

373 00:58:47.964 --> 00:58:52.644 This year, our big focus is moving from, from, um, to code first, Git native.

374 00:58:52.964 --> 00:59:05.363 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?

375 00:59:05.524 --> 00:59:14.204 This exists today- Okay... if you use the RC LI. So it's a virtual file system Uh, like, it's a representation of... It's like infrastructure as code. Okay. Yeah.

376 00:59:14.424 --> 00:59:24.164 And in fact, what we're announcing tomorrow is, like, we've done our kind of final change for declarative schemas. This was a, a really challenging one actually, covering all of Postgres.

377 00:59:24.684 --> 00:59:32.184 And so now you can have a full file representation of Postgres, uh, the shape of your database declaratively in your folder.

378 00:59:32.484 --> 00:59:42.524 That's really important for agents, 'cause generally how you operate with Postgres is you have a migration file every time you make a change. You add a file to this migrations folder, and it's like a historical log.

379 00:59:42.984 --> 00:59:52.684 Okay, so that's great. It's industry standard. But imagine putting an agent across a thousand files and telling, asking it, "Oh, what is the shape of this table?"

380 00:59:52.784 --> 01:00:01.724 It has to guess through 1,000 files, so- It has to recompile basically- It has to recompile... edit weights, yeah. Or you have to launch the database, um, locally- Migrate, migrate, migrate, migrate...

381 01:00:01.764 --> 01:00:06.304 run the migrations up, and then you inspect, introspect the database. But that's too slow.

382 01:00:06.404 --> 01:00:15.844 So basically what we want to do is turn the schemas folder inside our, our Supabase folder into a completely declarative state of your database.

383 01:00:15.884 --> 01:00:26.524 And so we've captured everything from tables, policies, uh, database comments, all inside these files. So we're releasing a new library, pgdelta. You can use it with any Postgres database.

384 01:00:26.804 --> 01:00:37.704 Um, but we've got mechanisms for the, for Supabase. And so this is an important step for agents. We know that agents are gonna operate mostly on code. They don't wanna reach out to anything.

385 01:00:37.744 --> 01:00:45.684 They've gotta iterate very fast on the file system. So that's fine, and today people are using CLIs. They're happy with these harnesses.

386 01:00:46.104 --> 01:00:56.044 What I suspect we're moving more towards is, like, at the moment I'm trying to do some development just on my mobile phone, for example, and I'm quite comfortably able to do it.

387 01:00:56.104 --> 01:01:02.764 Of course, I know what's going on behind the scenes, so I'm telling it things that I know it has to do from an infrastructure point of view.

388 01:01:03.064 --> 01:01:13.524 But we're getting to a state where I can imagine you don't need to be an engineer. You, you have an idea, you say to your phone, your dot or whatever personal agent you have- [chuckles]...

389 01:01:13.524 --> 01:01:21.684 "I want to build X," and one second later, you kind of have this application live. You can iterate on, on it yourself.

390 01:01:22.044 --> 01:01:32.364 So these are the things that I'm thinking, "Okay, what are the use cases that people are going to want if this is the case? Is it personalized? What do we care about? Are these things gonna scale up?

391 01:01:32.804 --> 01:01:40.084 What are the mechanisms? How are we going to develop, and what tooling do we need to provide to make sure that we are the default provider in this new world?"

392 01:01:40.644 --> 01:01:48.564 Any questions that you're, you're interested in, uh, as for a crystal ball into the future? As someone who's been a software engineer now for, like, a few decades,

393 01:01:49.464 --> 01:01:54.024 you have strong opinions on, like, how the software should be built, right?

394 01:01:54.504 --> 01:02:06.244 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...

395 01:02:06.604 --> 01:02:19.524 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.

396 01:02:19.564 --> 01:02:21.944 They're all gonna use Supabases. Fantastic. [laughs] Yeah.

397 01:02:21.964 --> 01:02:32.744 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.

398 01:02:33.304 --> 01:02:37.024 Like, are we gonna lose that or- Yeah, so you want diversity in ideas- Yeah, exactly...

399 01:02:37.064 --> 01:02:43.144 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.

400 01:02:43.404 --> 01:02:52.484 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.

401 01:02:52.604 --> 01:02:55.824 So first of all, I think that's a valid fear. I do think it's out of your control.

402 01:02:55.824 --> 01:03:06.344 [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.

403 01:03:06.904 --> 01:03:17.044 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.

404 01:03:17.064 --> 01:03:26.424 [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...

405 01:03:26.464 --> 01:03:32.944 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?

406 01:03:32.984 --> 01:03:39.684 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.

407 01:03:40.264 --> 01:03:49.604 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.

408 01:03:49.884 --> 01:03:59.244 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

409 01:04:00.164 --> 01:04:10.044 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...

410 01:04:10.104 --> 01:04:12.364 on, on the pod, so, you know, I wanna capture this.

411 01:04:12.944 --> 01:04:25.544 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...

412 01:04:25.584 --> 01:04:36.304 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?

413 01:04:36.304 --> 01:04:46.224 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.

414 01:04:46.224 --> 01:04:58.104 [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...

415 01:04:58.684 --> 01:05:05.964 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.

416 01:05:05.984 --> 01:05:12.524 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.

417 01:05:12.624 --> 01:05:23.004 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?

418 01:05:23.684 --> 01:05:33.804 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.

419 01:05:33.844 --> 01:05:45.464 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.

420 01:05:45.484 --> 01:05:53.904 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?

421 01:05:54.284 --> 01:06:12.064 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.

422 01:06:12.364 --> 01:06:20.544 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?

423 01:06:20.984 --> 01:06:29.664 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.

424 01:06:29.684 --> 01:06:42.184 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...

425 01:06:42.244 --> 01:06:50.924 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."

426 01:06:51.144 --> 01:07:00.952 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."

427 01:07:00.972 --> 01:07:08.912 [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.

428 01:07:08.972 --> 01:07:14.812 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...

429 01:07:15.252 --> 01:07:23.332 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.

430 01:07:23.972 --> 01:07:30.412 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.

431 01:07:30.452 --> 01:07:40.552 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.

432 01:07:41.092 --> 01:07:51.392 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.

433 01:07:51.732 --> 01:08:00.952 But two, also your dependencies are now becoming, uh, optional. Like, you can just vibe code whatever you want. [laughs] Anything, yeah. Yeah, yeah.

434 01:08:00.992 --> 01:08:11.712 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.

435 01:08:12.552 --> 01:08:20.532 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?

436 01:08:20.572 --> 01:08:27.052 Are you gonna rewrite Postgres?" Postgres is not rewritten and, uh, is not in Rust, but I doubt down to that level you care.

437 01:08:27.112 --> 01:08:38.352 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.

438 01:08:38.372 --> 01:08:53.272 [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.

439 01:08:53.632 --> 01:09:05.972 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.

440 01:09:06.472 --> 01:09:18.112 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.

441 01:09:18.512 --> 01:09:29.972 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.

442 01:09:30.352 --> 01:09:37.492 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.

443 01:09:38.012 --> 01:09:48.292 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.

444 01:09:48.792 --> 01:09:59.532 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.

445 01:09:59.912 --> 01:10:07.412 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.

446 01:10:07.792 --> 01:10:17.912 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.

447 01:10:18.432 --> 01:10:28.292 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.

448 01:10:28.332 --> 01:10:38.172 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.

449 01:10:38.212 --> 01:10:47.772 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.

450 01:10:47.832 --> 01:10:59.652 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.

451 01:10:59.732 --> 01:11:13.751 [laughs] So we have, like, the auth product, and we've always been very insistent that they should make their own decisions, like, on what tools they use, like, how they behave as a team, like, the reporting structure, and the AI policy was kinda no different.

452 01:11:14.212 --> 01:11:25.612 It's like a lot of people asking, "What is the AI coding policy?" And we were like, "Well, just it's up to you as, as your team." I think we have started to consolidate a few bits, especially around cost.

453 01:11:26.172 --> 01:11:34.252 But as much as we can, we'd like to keep that model of, like, let the teams decide how they wanna run things. That's it, very decentralized. Mm.

454 01:11:34.272 --> 01:11:46.152 [laughs] I, I would say, um, I am trying to figure out a number for company leaders to take home with them as, like, what is the market standard. And so, uh, what, what are data points you can hang your head on here?

455 01:11:46.492 --> 01:11:53.432 People say, uh, you know, the average, uh, person is spending, like, $20 a month, and then, uh, average employee is maybe $200 a month.

456 01:11:53.992 --> 01:12:04.752 OpenAI, uh, you know, at the, the sort of 50th percentile is spending maybe three to five thousand a month, and the top percentile is in the maybe hundreds of thousands. Mm. Right? And so that is the rough idea.

457 01:12:05.232 --> 01:12:15.832 I think, uh, a good metric is basically what is your ratio of token spend to human spend, and it's r- like, for a lot of people, it's maybe 0.1, uh, tokens to, to one human salary.

458 01:12:16.432 --> 01:12:23.192 And, uh, you know, you can probably go up to one pretty safely, and if you're over 1.5, you're probably in trouble. [laughs] Something like that.

459 01:12:23.792 --> 01:12:33.592 So I'll, I just, uh, say, uh, luckily, as a remote company, basically everything that's- Documented, yes... ever happened is documented. [laughs] So I can ask our AI how much we're spending.

460 01:12:33.672 --> 01:12:43.472 So currently, we spend about 300K per month. Yeah, uh, for 600 people. No, for- So like 450 people... 400. Four, 450 people. Yeah. Okay. Yeah. That, it gives people an idea.

461 01:12:43.832 --> 01:12:53.972 But, but what you've said about, like, scaling the cost in terms of related to the number of employees makes sense also for us because we are hiring very quickly.

462 01:12:54.192 --> 01:13:01.968 So if we end up getting a lot of leverage out of just, like, actual spend- You should be, right?... then yeah. You, you need a whole bunch of serverless employees that can spin up and down as, as you need, right?

463 01:13:02.008 --> 01:13:09.388 Yeah, exactly. Wh- which is like, yeah. Like, you know, I, I just had a CTO of, uh, you know, a- another top startup say, like, he was worried about his company spend.

464 01:13:09.768 --> 01:13:14.828 And I'm like, "Well, actually your- that's your investment in everyone else, and probably is very skewed," right?

465 01:13:14.848 --> 01:13:22.868 Like there's, there's like two or three people that are super insane at your company, and everyone else is like probably not using AI enough. Yeah, yeah. Yeah. [laughs] That's exactly what we see as well.

466 01:13:22.888 --> 01:13:30.128 [laughs] We've got a leaderboard, and people... We put the leaderboard up, and usually people are a little bit ashamed, so they kind of like- [laughs] Yeah, 'cause you're wasting money probably. Yeah.

467 01:13:30.148 --> 01:13:37.528 [laughs] And that's also the wrong focus. Like, the focus shouldn't be on how much you're spending, but, like, how much end-to-end... Like, in terms of systems thinking- Yes...

468 01:13:37.548 --> 01:13:50.388 you should be thinking, like, in terms of how many end-to-end processes are you actually solving that where you no longer need a human in the loop, or collapsing down long-running human problems that suddenly become very short problems.

469 01:13:50.448 --> 01:14:01.588 So that's where we're spending most of our investment on, like, how we're using AI. Luckily, you know, as you saw, like, every single thing that's ever happened in Supabase is documented because we're an async company.

470 01:14:01.888 --> 01:14:10.888 Yeah. So building a company harness or, like, an agentic harness around Supabase is basically just formalizing. You can point AI at it and say, "What are our principles?"

471 01:14:11.168 --> 01:14:18.788 And it'll docu- d- document down all our decision-making principles for the whole history of Supabase. So that's where we're spending a lot of our time.

472 01:14:19.148 --> 01:14:30.008 Before we move on from people, we do have, I think, now, like, 50 open roles. [laughs] Yeah. And we are fully globally remote, so we don't care where you are. If you're very good- Same, same salary everywhere, right?

473 01:14:30.068 --> 01:14:36.668 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...

474 01:14:36.768 --> 01:14:47.908 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.

475 01:14:48.248 --> 01:14:55.688 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?

476 01:14:56.348 --> 01:15:05.088 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.

477 01:15:05.148 --> 01:15:17.288 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.

478 01:15:17.328 --> 01:15:25.428 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...

479 01:15:25.468 --> 01:15:33.148 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.

480 01:15:33.388 --> 01:15:39.168 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...

481 01:15:39.328 --> 01:15:50.328 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...

482 01:15:50.548 --> 01:16:03.388 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...

483 01:16:03.468 --> 01:16:11.208 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...

484 01:16:11.308 --> 01:16:19.828 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.

485 01:16:19.908 --> 01:16:28.748 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.

486 01:16:28.748 --> 01:16:38.108 Thanks, thanks for having us. [upbeat music]