0:00
I think the thing that keeps you grounded is just absolutely knowing that money will not solve your problems.
Search this video
0:03
And that's a good thing for for startups as well, like early stage startups.
0:07
Like we should not exist if AWS are so good.
0:13
They've got way more cash than us, but we've been able to for many different reasons.
0:20
>> For this next session, I'm welcoming uh Paul Copplestone.
0:23
Is that the right pronunciation? Yep. >> Okay, awesome.
0:27
>> Who is the co-founder and CEO of Supabase.
0:29
Uh Supabase began in 2020 as an open source tool uh to deliver Postgres, right?
0:36
Like help people host Postgres online.
0:39
Uh today, it's one of the fastest-growing dev tool companies in the world, if not the biggest.
0:48
Maybe well, you're not going to uh say no. Okay.
0:52
Uh they just became a decacorn just a few years after founding, which is insane.
0:57
Uh please welcome Paul Copplestone.
0:59
Great to have you with us.
0:59
Uh I really love dev tools, as you know.
1:01
And so it's kind of like having the chance to uh ask you questions is always great.
1:05
Um but before building uh Supabase, I mean, you started with like was that your third startup? >> Yeah, that's right.
1:14
Um so like my the short history getting to where we are today >> Sure. Let's go for it.
1:21
>> is uh well, going right back, my father's an entrepreneur, and so it's kind of in my blood, and I've only really done >> Where are you from?
1:29
>> Uh I'm from New Zealand. Yeah, originally.
1:31
>> Not in Europe, that's the only one. >> Yeah, yeah, yeah.
1:33
>> Just living in Europe right now.
1:34
>> I right now I spend most of my time here in Europe, so my French is not so great, unfortunately, but I'm practicing.
1:39
So yeah, literally the other side of the world I grew up, and then my father was an entrepreneur, so I very much felt like a natural thing to me.
1:49
Then I did one job in a corporate uh consultancy, and >> You loved it.
1:55
>> and I absolutely hated it.
1:55
And uh since then I've only done kind of startups or entrepreneurship.
2:00
So I yeah, this now officially is my third venture backed startup where I've taken VC funding, but I've actually done quite a few in the meantime.
2:13
>> What happened to VC's money for the first two?
2:16
>> Just blew it on uh Yeah, having great parties.
2:20
No, I think like the I don't know is this the recorder?
2:23
I was going to say something much more vulgar, but uh No, we I mean I think it's the number one reason why startups fail is they run out of money, isn't it? >> Not really. >> No, is it?
2:40
>> Well, actually look at the numbers.
2:41
>> Yeah, where were was the numbers?
2:42
>> Uh sorry, founders giving up. >> Uh okay, okay.
2:45
>> Drama, not finding product market fit.
2:48
And we have so many companies, I mean I'm speaking about the YC companies on the data on YC companies. >> Yeah.
2:54
>> Um people usually close their companies they still have a lot of money. >> Okay, okay.
2:59
>> So no, maybe your case is different.
3:02
>> Yeah, well, I mean I guess we didn't run out of money, but eventually if you are persistent enough and you can't find product market fit the inevitable end cases that you run out of money.
3:13
So the first startup I was the CTO rather than the CEO.
3:15
We didn't really find a product market fit and I left after a couple of years because the way that I wanted to take the business was different from how my CEO wanted to take the business.
3:27
So much as you say that's reason number one. Yeah.
3:31
Reason startup number two was very different.
3:33
In fact, I had said to my co-founder the co-founder now, I said to him, I'm going to do a dev tools company and he said to me, well, just give me two years.
3:44
You be the CTO, help me launch this business cuz we had worked together before and I said that's fine.
3:49
I'll be incubating Supabase."
3:54
And then I helped them launch the business and when it felt like the right time, then I actually launched out of it.
3:58
So, I still that company is still going and is profitable.
4:02
But um it wasn't, you know, for me.
4:02
It was never going to be my the thing that I work the rest of my life on.
4:09
Whereas now Supabase is is very different.
4:12
>> I didn't realize that.
4:12
So, when you launched Supabase in 2020, you were actual you'd been working on that for 2 years as a side project?
4:20
>> Not like I wasn't sure.
4:20
I knew I wanted to do a dev tools company, but I wasn't sure exactly how it would function.
4:24
So, um and I also knew I wanted to do a database company.
4:29
But the sequence of events was um basically I was using um Postgres, which is what we use now.
4:36
Um I'm not sure how technical the audience so I'll just keep it kind of high level, I guess. Pretty technical, okay.
4:42
So, uh Postgres and Firebase and I was using both of these tools in my previous company and then when we um hit some scaling limits on the Firebase side, I decided to migrate it across to Postgres and I wrote an open source tool.
4:57
I open sourced it, put it on Hacker News.
4:59
And um then it started getting some traction in the open source.
5:05
But there was only maybe 3 or 4 months before I launched Supabase.
5:09
So, it just felt the right time to build the database startup that I wanted to build.
5:13
And so, um I was trying to figure out how to do it and that was just kind of the impetus. >> Why you Postgres?
5:20
And Postgres was already already what?
5:21
30 years old or something?
5:24
>> Yeah, Postgres was Postgres well, in its earlier forms it's about 40 years old.
5:28
But um there's a few things that I saw.
5:30
Um I definitely knew I wanted to do an open source company.
5:35
Uh I knew I wanted to do a database company.
5:37
I used to consume Hacker News all the time.
5:40
So, for the developer segment, I kind of knew what was up and coming.
5:46
>> How many people here know what Hacker News is? Okay, a majority.
5:51
>> Yeah, I I probably read it like three times a day and uh show Hacker News, I monitor them.
5:55
So, you kind of get this sixth sense of what is trending and I could see that Postgres was trending.
6:00
It wasn't the most popular database at the time, but um now it is. So, it was a good bet.
6:06
And the other thing that I really liked about it is if you are going to build a database company, you have to have a good reputation because to build a database is incredibly difficult.
6:15
But um Postgres kind of had that reputation built into it and it had another thing that was really important.
6:23
When you think about um which database will win over the next 50 years, it's a little bit like game theory where I will wager that Postgres will continue to be the dominant player because no one owns Postgres, which means that everyone contributes to Postgres.
6:42
Then all the hyperscalers have to have the best Postgres offering because it's the most popular and therefore Postgres continues to to improve.
6:49
So, it's this flywheel effect that comes in and I really liked that dynamic behind it and we tried to lean into it to to win that space.
6:59
>> Firebase was not Postgres back to write down.
7:01
>> No, Firebase was more I think it started on and then eventually moved off. Yeah.
7:05
>> And in the way you were started because you started open source too.
7:07
I remember seeing you as like kind of the open source Firebase.
7:11
>> Yeah, we we did some positioning.
7:11
So, after I wrote that real-time engine uh and put it on Hacker News, I thought that was the thing.
7:18
Uh everyone thought real-time was cool on top of Postgres.
7:21
So, I wrote real-time Postgres and it actually didn't really take off and then we switched the tagline to open source Firebase alternative.
7:28
And then from that point on it really took off.
7:33
>> So, this was your early marketing. Um why open source?
7:35
What made you decide to be open source?
7:38
>> Yeah, I think this one's more just philosophically. I like open source. I like that community.
7:42
I like building communities.
7:45
And so, you know, a lot of people will think of them perhaps as a tactic or a strategy, but in this case, there wasn't much of a discussion.
7:53
Both my co-founder and I like open source.
7:56
>> So, you just like did it so you wanted to do it. >> Yeah.
7:58
>> Um were you thinking about monetization from the get-go or >> Yes. Yes, we were.
8:03
Although, it took us about a year and a half to start earning our first dollar, but we knew that databases were super interesting because they're incredibly difficult to get a motor to break into that market.
8:15
But once you do, then they grow. They grow naturally.
8:19
You You founders, if you start a business and your business is growing well, you go to sleep and your business continues to grow.
8:25
The thing underneath that is a database.
8:27
So, you know, we like that model.
8:31
>> So, you saw a lot of expansion of usage from small startup would start using you because you were easy and then would go with you and become good customers.
8:41
>> Yeah, and our initial customer base, this is why we applied for YC, was high-growth startups.
8:45
And so, the ideal ICP was someone who was venture-backed and was likely to grow fast who happens to be a Y Combinator company.
8:53
>> How many YC companies are using you?
8:54
>> So, we get about 60 over 60% of every batch will use Superbase. >> Oh, okay. Including yours?
9:01
>> We Yeah, we use Superbase to build Superbase as well.
9:03
>> No, but I mean including your batch.
9:06
>> Uh include Uh no, our batch was um >> [laughter] >> No one No one trusted us at the start.
9:09
I think there was zero uh Well, there was one company.
9:15
>> But uh like a lot of companies trusted us during our batch. >> Did they? >> Yeah. Yeah. Yeah.
9:18
Like I think >> They're better than us.
9:19
We're not very trustworthy people, naturally, so.
9:23
>> Well, your batch was remote, too. I didn't tell. >> Yes.
9:25
Yes, that's the other thing.
9:25
We couldn't really network with anyone, so we were just going straight out to the internet. >> That's awesome.
9:30
Um I mean, open source when uh when we're I discovered what you were doing there, I was actually impressed because you are fully open source open source like 100% open source.
9:40
There is no like enterprise version that you charge for.
9:45
>> Yeah, so there's different >> a risky choice?
9:47
>> Yeah, there's a few models that you can run if you run open source, which is like a a license where you might put some of it behind uh flag and you're going to pay or enterprises have to reach out.
9:56
For our model, we give everything away.
9:58
You can use it completely free.
9:59
The only thing that's not available is like how to host a million databases.
10:04
Mainly because that code is just ridiculous spaghetti.
10:07
Like you can't actually really push it out in an easy way.
10:11
So >> So that code is not >> That code is not available.
10:14
It's like lumped in with billing logic.
10:16
It'd be a security nightmare for us to put it out.
10:20
>> And what uh prevents any kind of like big cloud provider to actually host Firebase themselves and resell it?
10:27
>> Well, >> They've tried. >> They've tried.
10:30
>> [laughter] >> Yeah, I mean, of course like um you know, you'll see examples where they might take Redis or Elastic Search or something like this.
10:39
>> they did that risk before or like like >> Exactly.
10:40
And then they'll host that and they'll actually make more revenue from that than the companies that created them.
10:48
For Supabase, in some ways we did the opposite because they host Postgres databases and we just started counter-positioning against RDS and um and Aurora.
10:59
And so, you know, they already had these offerings.
11:01
What are they going to take?
11:03
And then the other thing is it's Supabase is actually a stack of tools all working well together.
11:09
For them to do this, it's not the same as hosting a Redis instance.
11:14
It's actually incredibly difficult to manage all these different tools together.
11:19
>> And how about direction of the community today then?
11:22
Uh I remember you you said that you always have people complaining that you are not what?
11:26
Open source enough or helping them enough? >> Yeah. Yeah, yeah, yeah.
11:30
But that's always going to be the case, right?
11:31
I mean, you always find customers and they're always the ones that um are loudest.
11:38
Are the customers like it can be open source community members and you might not be open source enough or you haven't released enough um or it could be that your pricing like your customers your pricing's too expensive.
11:49
Um there's always going to be that crowd.
11:52
You can you can't please everyone.
11:54
That's just >> ignore them. >> Yeah.
11:56
Or well, listen to them ideally because if um but you have to still make a judgment on whether it makes sense what they're asking or not.
12:04
>> So uh nobody wanted to use you to win your badge.
12:06
How long did that take you to win the trust of developers?
12:11
>> Um >> And you are like Firebase was still in the equation, right?
12:13
People were still using Firebase. >> Yes. Yes, they were.
12:16
I think people loved the idea of Supabase from the start and so they really loved the positioning Postgres open source.
12:23
And they knew they also loved Firebase cuz Firebase is legitimately a good tool.
12:29
It just lacks Postgres and open source.
12:31
So the idea that you could have the same experience but with these two elements was very appealing.
12:38
Then all they needed was the trust that we could host these sites.
12:43
So as we started building up more and more credibility with larger startups um building with us then maybe yeah, as I said, we started monetizing in the first year.
12:53
But really it was kind of still slow.
12:55
I think the first year we exited with revenue in the order of a million ARR or something like that and then like the next year seven and a half.
13:05
So it was slow going but then it just took >> Well, I think a lot of people would love to have like go from one to seven in a year. >> That's true. That's true. Yeah.
13:14
>> That's slow going compared to today.
13:16
>> Well, today people go to a hundred million in a month.
13:18
So >> Well, not that many.
13:22
So yeah, uh let's be careful about what you read in the press because on you only read about the crazy outliers.
13:28
Most companies don't go from zero to 100 in a in a year.
13:32
>> Yeah, that's also true.
13:33
>> I've always been impressed by the community behind Superbase and like how you got them to follow you like the developer experience created.
13:40
Like how do you think about these things?
13:44
>> Yeah, well, when it comes to community, we don't worry too much about getting them to follow us, but we do care about that second part, developer experience.
13:53
And how do you get a good developer experience?
13:55
There's a couple of elements.
13:56
First, you need to know what developers want, so you need to listen to them.
13:59
And so much like the YC motto, just like figure out what your customers want.
14:04
And for us, that is the community.
14:07
We'll just poll the community all the time finding out what they want, listening, seeing where they are.
14:10
And your community will ideally live anywhere.
14:14
So we're on Reddit, we're on Twitter, we're on Hacker News, wherever we need to be to find the the developers.
14:22
And then for us, we really thought hard about developer experience at the start.
14:27
The key thing that we measured, I remember right at the start, was time to value.
14:31
So a a customer I remember I actually timed how long it took to launch an RDS instance on AWS and then connect it to and then insert a row.
14:41
And it was something like 8 and 1/2 minutes.
14:43
And we decided, well, we have to get that at least under a minute for us to get time to value.
14:50
And then we just started tracking that as soon as fast as we can.
14:54
And >> these days it's 5 seconds or so. >> now. Okay.
14:56
And how long is RDS today?
15:00
>> Probably still 8 and 1/2 minutes, yeah.
15:02
>> Did you actually communicate on that?
15:04
Was that part of the >> No, we didn't say it out loud.
15:05
It just kind of innately becomes a part of the experience of the developer.
15:09
They don't want to like when you want time to value, you want these days it would be an agent, right?
15:16
Imagine you just kick off an agent and you say, "I'll build me an application."
15:19
If you come back after these days you can walk away from your computer, but if you come back after a minute and it hasn't solved the problem and you need to re-prompt it, that's actually a loss of value.
15:29
So, what we want to be able to do is really one shot.
15:32
That So, it's no longer a time element.
15:34
It's the ability to get you to your result as fast as not just as fast as possible, but with as little input as possible.
15:44
>> Uh let's speak about that. Like AI uh shift here.
15:48
It feels like everything we've discussed so far is kind of the first life of the company and now you have like a second life that was kind of triggered by all of these AI use cases.
15:57
Is that what Is that right?
15:57
Uh so uh it's kind of like you became the default back end of pretty much every AI application started with uh the lovable of the world, right?
16:09
>> Yeah, we in 20 In the December 24, we saw we're getting DDoS'd by uh a bunch of companies which were launching all these applications, databases, and um the two main ones were Bolt and Lovable.
16:22
And um and now like fast forward to today, compounding Claude code and Codex, I think it's now that we can measure 60% of databases get launched by an agent, but more likely it's like 90% and it's in the millions every every month.
16:42
>> When did you realize that uh these AI use cases were actually a big thing versus like uh like what you thought you would do? >> Well, we don't.
16:50
This is kind of the thing where you look back and you think, "Oh, it's super clear.
16:54
Of course, um we should do this."
16:56
But, when we went back, we looked, for example, at Lovable when we tried it and the apps that it was building back then cuz it was just getting started were were not that great and we thought, "Oh, you know, do we >> Have great customers. >> No Well, >> I know.
17:10
Eventually probably, but like >> Yeah, exactly.
17:14
And we have this thesis, you know, you can launch these but will they grow?
17:17
And the will they grow is kind of the thing that we care about.
17:22
So, um there was a lot of debate internally, oh, you know, do we even want to support this this use case?
17:26
Or or as well, people were saying, do we want to build our own lovable so that we have access to the distribution?
17:33
But now it became very clear our product that we built was Superbase for platforms, and a lot of companies, I think it's over 50 companies built on top of this product where they can launch millions of databases.
17:48
>> So, the Superbase for platform is a product created because of the success with lovable and Bolt. >> Exactly.
17:54
We we kept iterating, it was pretty close to where we were, and we kept iterating on it, and then we realized, oh, this is a good product.
18:01
Like we will launch it for everyone to get started.
18:04
And the other thing that made it quite easy is that if you squint your eyes a little bit, this is what enterprise customers want, especially innovation labs.
18:12
They want the ability now to launch many databases in their own company, and have the ability to see which ones are unsafe, uh security on top of them, shut them down if they're growing too fast, or you know, manage them.
18:27
So, we could see that we can merge these road maps somehow. >> All right.
18:31
So, from the lovable of the world that have like millions and millions of instances, then you also have like the enterprises who wants to manage fewer number but bigger databases. >> Yes, exactly.
18:42
They and it could be in the millions.
18:45
I I don't know if you guys all saw the Snap specs, um where meming the internet that the back end for them is Superbase as well.
18:53
So, we knew and larger enterprise customers were were wanting these things as well. >> That's cool.
18:59
And uh I mean uh that was kind of like the vibe coding era.
19:05
It feels like things shifted again last year.
19:09
>> Yeah, so then Christmas, um we came back from Christmas and everything was going up into the right again and accelerating, which is crazy cuz I think by that stage we're at maybe 6.
19:18
5 million developers and today we're at 10.
19:24
So, we've already layered on what 4 million >> Remember last year you were on stage in Paris. >> Yeah.
19:30
>> Your numbers were insane. >> But yeah.
19:33
>> But it was nothing compared to today. >> To today, yeah.
19:35
>> How much like like since last year more like 10x? More than that?
19:39
>> Yeah, probably something like that.
19:41
Which is insane off the base.
19:41
I think that's the crazy thing.
19:43
Like AI kind of shifted what normally happens is you might pile in because we don't we don't spend money on marketing.
19:51
Like I I tell people if I could give a free database, I'd rather give that to developers than, you know, spending on marketing.
19:57
But usually what you do is at our size you'll pour some money into marketing, which means that the intent of the users plummets a little bit.
20:05
So, you start seeing your conversion rates go down or your activation rates go down.
20:09
But actually what we saw was basically every chart started going up into the right.
20:13
So, the number of users was accelerating while conversion was going up and activation was going up.
20:19
Just everything was going up. >> It was cloud code. >> Cuz of cloud code.
20:24
>> So, people started to build like to pretty much build more and build bigger applications that they would do on with pipe coding simply.
20:32
>> Bigger, more successfully, they could get to a product sooner.
20:34
They were building more products. Yeah, it was incredible.
20:39
>> Like in the first you make more money from more usage, I guess.
20:42
>> Yeah, it's all usage based. >> That's incredible.
20:45
What does that mean for like we're speaking of developer experience?
20:46
Uh what does that mean for you for your developer experience that agents have become the majority of your actual users?
20:56
>> Yeah, there's a part of it where we went through Actually, we went through kind of three phases.
21:01
At the start, um we built a dashboard and the dashboard even had like click around to add things to your database because there was a discovery mechanism.
21:13
Then when the LLMs came out, we kept the dashboard but we stripped out a lot of the panels and it just became code SQL editor and chat interface.
21:26
Now of course when agents don't give a about any of those two things, they just care about CLIs and MZPs.
21:30
But what we try to do in terms of developer experiences get this unified experience and where things are naturally going to go is ideally the entire experience that you would have in your dashboard, you would also have in your code.
21:43
So for example, a representation of your database in code in declarative schemas.
21:51
Basically it becomes infrastructure as code inside the Superbase folder.
21:53
And then what spills out the other side of when you've got it in code, it has to be in Git.
22:01
What spills out of that is when it's in Git, you have branching, so many environments.
22:06
So basically the entire Superbase stack has to be branchable.
22:11
And this is basically where we're heading now where the agent has to be very good at spinning up an entire database from a user intent but then also managing it and developing on top of it as it gets more complex. >> Uh we're in 2026.
22:26
If you were to start today, I'm thinking about all these founders who may be thinking about building dev tools.
22:33
How differently would you do that?
22:35
Because you would target agents from the get-go, right? >> Oh yeah, for sure. For sure.
22:40
>> Any ideas of what people should try? >> Like what ideas?
22:44
>> What not not ideas what you should build, but any ideas how they should what they should do to make sure agents will find them and use them.
22:53
>> Oh yeah, that's a tough one.
22:55
Um I don't know, but when it becomes a race of like oh, um let's assume that agents will only choose three. And that's it.
23:07
That means, okay, there'll be three winners, no one else, maybe three of you win.
23:12
All the rest of you are competing on some other thing.
23:14
So, what is that some other thing?
23:16
Probably exactly the same as what has been in the past, right?
23:18
So, you should still be doing things to make the agent extremely easy to build with you, but like becoming the default for them is just going to be one of those rare rare occurrences, and you'll have to have a lot of luck.
23:31
I don't like betting on luck.
23:32
You just do the things that you have to do in order to get word of mouth, uh in order to um make it very easy for the experience, and then you just pray that you end up being one of those defaults. >> Right.
23:46
Being open source probably helped a lot for you.
23:49
>> Open source, for sure.
23:49
For us, uh I mean, we have a community, uh we have a lot of Reddit comments, we have a GitHub community, all the comments there.
23:56
The LLM slurped up a lot of >> Yeah.
24:00
What's that mean for, I don't know, like, from now on, like, developers don't even need to know about you? >> No, they don't, yeah.
24:07
>> They use you without even knowing who you are.
24:09
>> No, no, but you know, I think we talked about this once, like, there is this kind of idea that then eventually, if that's the case, um only AWS wins, and we might as well all pack it up and go home and like just admit defeat, but you don't really know, so you're just going to be in the arena kind of fighting it out.
24:29
And and you know, maybe it ends up completely different from from our hypothesis that there's only one default winner.
24:34
And I imagine that you know, people won't choose Supabase in the future because they just like our vibes, and they'll say, well, I prefer to choose Supabase over AWS.
24:44
And maybe if AWS can get their experience as good as ours, we won't have much of an advantage except for the vibes. >> Perhaps it is.
24:58
Um All right, let's shift back to uh fund raising.
25:01
So, I mean you had the crazy uh couple of years.
25:07
Your is like I was looking at the numbers like three incredible rounds, one more incredible than the last one uh each one.
25:12
And then the the the round you just raised like 500 million at over 10 billion valuation.
25:17
Uh how do you continue uh I actually running the company when all of that happens to you?
25:25
Like how do you do with the make sense of your life and uh and what you should do?
25:31
>> Uh um like personally?
25:34
>> Yeah, personally and then probably as a as a as a as a team. >> Uh okay, yeah.
25:40
Well, the money shouldn't change how you operate, in my opinion.
25:42
Like the you you have jobs to be done and the money is in service of the job to be done.
25:50
And if you are thinking, "Oh, now we've got this money, let's do something else."
25:53
then you're probably not going about it the right way, so >> What what keeps you grounded?
26:00
>> Um go to Buddhist monasteries.
26:02
>> [laughter] >> I as a as a company, I think the founders have to have um you can't get caught up on money won't make you succeed, basically.
26:14
And the founders have to really know that and there are some ways that it can help you succeed temporarily, but at the end of the day, if you want to build kind of a generational company, then you have to do a bunch of things correctly and so I think the thing that keeps you grounded is just absolutely knowing that money will not solve your problems.
26:33
And that's a good thing for for startups as well, like early stage startups, cuz it means like let's say you want to compete with Superbase, it means that we are limited not by the amount of cash that we have in the bank.
26:48
You can compete with us on many other vectors in the same way that we can get to where we are.
26:53
By by all rights, AWS should never have been able like we should not exist if AWS are so good.
27:02
They've got way more cash than us, but we've been able to for many different reasons.
27:05
So, the cash is really just in service of all those things that would allow us to win even if we didn't have the cash.
27:13
>> How about the relationship with your investors?
27:16
Does that gives them more power or less power?
27:20
>> Giving them ca- >> No, like growing like you did.
27:23
>> Ah, um >> You feel like some pressure, even stronger pressure, or raised expectations?
27:30
>> Yeah, the the more success Well, we had this funny dynamic, actually.
27:32
So, normally when you start off in a you you raise your initial seed round from a seed investor, we had a bit of a funny dynamic in that a hedge fund invested in our >> Right. >> Koto.
27:45
Hedge fund, in case you don't know, >> Oh, that was when they were investing anywhere. >> Yeah, exactly.
27:49
So, these are these guys have got like 70 billion under under management.
27:54
So, at the start they're like, "Oh, here's 6 million.
27:56
Like we don't care about you."
27:58
And then as we started getting more successful and it looks like we might go to the public markets, then suddenly they're very interested in us.
28:05
So, that's a bit of a different dynamic from what most seed investors, as you become more successful, the seed investors become less interested because they've made their return.
28:15
But the the last investors that you had are very interested in your how much you'll return for them. >> Okay.
28:24
>> A interesting dynamic.
28:24
Um So, you started in New Zealand I mean, you are from New Zealand.
28:28
You actually started from Singapore.
28:31
>> Yes, I'm playing from Singapore.
28:33
>> And build the teams that distributed You are all remote. How many countries?
28:37
>> We have people in over 60 countries. >> Okay. How many people? >> 360.
28:43
>> Okay, so like only like a few people a handful of people per country. >> Yes, exactly.
28:47
Yeah, completely no offices, so it's all online. Yeah.
28:52
>> How do you build a company like that way? >> Um >> Do it again? >> Sorry?
28:57
>> Would you do it that way or is that the only way you know?
28:58
>> I would absolutely do it again.
28:58
Um >> That's That's why you can travel to Paris and like >> Yeah, yeah, yeah, exactly. Be based here.
29:05
I think um you can't half-ass it for starters.
29:06
Like you can't do the hybrid model where you've got to have some people distributed, some people in an office.
29:14
For us, we do everything written online.
29:18
The reason why I'd do it again now is because especially with AI, um we're starting to kind of use AI in our business and it turns out that everything that has ever happened at Superbase has been written down on Slack and Notion.
29:29
And everything every decision that is happening is happening written down.
29:34
And so for us to say to an agent, "Hey, what was that decision that we made 6 years ago?"
29:39
Uh we can actually prompt it and it'll find it in Slack >> even though it was right at the start.
29:46
>> So most people have agents who know everything for the last 6 months. >> Yeah.
29:50
>> They know everything from the last 6 years. >> From the start.
29:52
So if I want to build the fundamental principles of how we operate in Superbase, the agents can do it.
29:59
>> Uh we're nearly out of time. Okay, last question.
30:02
Uh is there any bet uh that you are thinking about now that uh you would that would drive what uh the company's going to do in the next 5 10 years? >> For Superbase?
30:15
Yeah, I think um and this one might be interesting for for you as well.
30:17
A lot of the time the agents uh thinking right now like in the build stage, so prompting, "Hey, I need an application.
30:28
Uh can you build it for me?"
30:28
But there isn't a lot of emphasis and the much harder thing is like how can I continue to operate this once it's built.
30:34
So a database is really one of these problems.
30:39
So for us it'll would self-driving databases, how to operate the database so that you don't have to wake up at night and see oh my database is down or it's insecure or all these patches come and all these things should be self driving essentially.
30:52
So, I think this is also interesting for many people because what you see is a lot of these startups launching now that are very easy to replicate because they're in the boom stage. It's much harder.
31:05
You want to focus on hard problems, right?
31:06
And the operate stage is one of those harder areas to to actually launch in. >> Excellent. Thank you, Paul.
31:12
Big round of applause for Paul. >> [applause]