Startup Explore

Philippe Noël (00:00)
Today, Postgres is the second most deployed relational database behind MySQL, and it's the fastest growing one. So it's probably going to become the most deployed one within a few years, if not earlier.

our thesis was always, if you want to build something that's 10x better, you need to make a much different...

architecture decision. And so Parade has decided to go all in on Postgres.

I think there's a really huge coming to build here, like $20, $30 billion valuation company.

Ali | Startup Explore (00:33)
Meet Philip Noel, the co-founder of ParadeDB, a seed stage tech startup that is redefining PostgreSQL database in a truly unique and transformative way. Now, despite being early stage, they've already secured a few enterprise logos, including Alibaba. So let's dive in.

Startup Explore (00:51)
Philip, thank you so much for coming. Very excited to have you here.

Philippe Noël (00:55)
Likewise, thanks for having me.

Startup Explore (00:57)
All right, let's dive in. Can you tell me what you're building and what is that big problem that you're trying to solve?

Philippe Noël (01:03)
Yeah, I'm building a company called ParadeDB. You can look us up, paradeDB.com. We're building a competitor to a database tool called Elasticsearch. And we're doing that on top of a different database tool called Postgres. The short story for people that are not familiar is the database that powers most of the software applications that you see today isn't very good at doing, let's say, search-facing experiences. So you go on Amazon, you search for a product that actually uses a different database than the one that might be

let's say responsible for processing the business data of the company. And when you have two databases, you need to keep them synchronized. You need to send data between them. That opens up room for a lot of things to go around and ParadeDB solve this problem by essentially upskilling the database that most companies use, which is called Postgres and making it capable of rivaling with a database called Elasticsearch to build those search facing experiences.

Startup Explore (02:00)
Got it. So if I understand correctly, basically the traditional way of setup would be something about having the Postgres QL system or database, and then you layer additional systems applications on top, and then you would have like ETL tools in between, and that would complicate things versus what ParadeDB offers is now consolidated features around the search and advanced analytics, and it all lives within the Postgres.

Philippe Noël (02:15)
Right.

Correct. Right. That's a very good summary.

Startup Explore (02:30)
Awesome. Awesome. And then what would be some practical benefits to businesses going with ParadeDB, whether it's like reduced cloud costs or reduced number of specialized skills that they need to maintain those systems?

Philippe Noël (02:48)
Yeah, excuse me. There's three main reasons people pick us. The first one is when you have those ETL pipelines, as you mentioned, between two databases, it actually creates, it's very, it's responsible for lot of downtime for applications because they break for a variety of reasons. It can also be in a situation where the data that is in each of the two databases is different due to something that went wrong or some delays in the propagation. They can directly impact the user experience.

makes that go away. You don't need to do this data movement. And so that problem goes away. Another one is in terms of performance and cost, Parade is a lot more performant and a lot cheaper. This is due to the fact that the data representation for these two databases is different. And when you convert one to the other, you incur compute. But you also can't make use of all the same optimizations that are specific to something like Postgres.

And so this leads to more data being accumulated on the search database. And then the third reason is Postgres is known for being a very reliable database and it has very high data integrity. A database like Elasticsearch is known for quite the opposite actually. So for a lot of companies that have very high requirements, like in the financial sector, for example, that reliability is very, important. Those are typically the main reasons people pick us. You're right though that in...

in practice, more people are familiar with Postgres. And so there's some sort of nice benefits to making it easier to have skilled employees that can man, you know, the systems, but those are kind of, I would say the side benefits.

Startup Explore (04:26)
Got it, got it. Yeah, no, those are pretty strong points. Now, let's talk about some of the use cases. And the most exciting use case that I saw was with Alibaba. And you guys being such a small company, early stage tech startup, how were you able to get on Alibaba's radar?

Philippe Noël (04:45)
That's a good question. So they reached out to us, Alibaba integrated our product within one of their data warehouses, which is for people unfamiliar, it's basically like really large scale databases that are used for processing a lot of data asynchronously. If you want to do this kind of functionality in Postgres, there's no tooling other than ours. That's it. There's some like basic functionalities within Postgres, but they're not anywhere close to what a dedicated database offers.

And in Alibaba's case, going to something like an elastic search would have been really, complicated for a variety of reasons. And so they basically had two options. Find us online. I actually don't know how they found us. I mean, we're an open source company. I'm sure they their research that way. Or they could build it from scratch. And they decided to purchase our solution.

Startup Explore (05:35)
Got it, got it. Yeah, I think probably having this online community helps as well, especially if they are supportive. But again, that's probably a testament to the quality of your product that you're building. Cool. And then during the evaluation process, I believe Alibaba actually compared you to Elasticsearch on their final stages. So what were some of the benchmarking or performance metrics that they used?

to make the decision.

Philippe Noël (06:06)
Yeah, they usually submitted. Yeah, the results were quite nice. They compared us in terms of performance to index the data, which is the process of making it searchable to the application. They also did benchmarking around the amount of search queries they could run concurrently. For example, they found parade to be about the same size of indexing, about twice as fast in the processing of the index, and then about five times faster.

Startup Explore (06:30)
Nice.

Philippe Noël (06:35)
in the query per second. so they were honestly, those results were better than expected. So it was not nice.

Startup Explore (06:42)
Yeah.

That was actually my next question. wanted to ask you, were there any big surprises for you? I'm sure Ali Baba was surprised to see those results, but were there big surprises for you during the evaluation process? And it seems like there were.

Philippe Noël (07:01)
Yeah, there

were. were. mean, we were also much earlier. So when they reached out to us, this was less than a year ago, maybe around April, but it's coming close to a year. And obviously, you know, by now the product is very good. But about a year ago, we were less than a year old as a company. It worked well, but there were a lot of things that needed improving. We certainly didn't expect it to perform that well.

Startup Explore (07:24)
Got it, got it. Now with Alibaba, I know it's a pretty huge enterprise company. have vast amounts of storage and data that they need to perform analytics over and do search. That being said though, do you see ParadeDB benefiting only like medium to large companies or you think smaller companies, including tech startups can actually benefit by using ParadeDB?

Philippe Noël (07:52)
Yeah, we have a lot of smaller organizations that use our product. We're open source, which means you don't need to pay us to be able to use it. A lot of small organizations will benefit from the free version. Most of the revenue that we make in the areas where we do things that, know, yeah, the revenue we make, basically the add-ons on top of the free version appeal more to the big enterprises. That's why we talk about them more often. We have partnerships that are going to be coming as well with

Startup Explore (07:59)
Mm-hmm.

Philippe Noël (08:21)
database tools that are very popular amongst small startups, where Parade will be available directly within there. And so I expect that will also facilitate adoption for smaller companies.

Startup Explore (08:32)
Got it. Got it. Cool. And then how you guys were able to basically extend the native capabilities of PostgreSQL in those two areas, like real-time search as well as the analytics. What were some of the big challenges that you faced along the way and how were you able to overcome those?

Philippe Noël (08:52)
Yeah, that's a good question as well. I mean, I don't know how technical the audience is or how curious people are to go into those details. yeah, yeah. But at a very high level, there are two things that we do that people build in databases to make them faster for performing search and analytics type of queries. One is to basically process the textual data to store it into

Startup Explore (09:01)
You can keep probably at the higher level.

Philippe Noël (09:21)
tokens or like chunks of words essentially that can be used to match against. And Postgres did not have the capabilities to do this with the state of the art algorithm that exists. They implement a more outdated algorithm. And so Parade has introduced these algorithms inside Postgres. And there's a lot of APIs that can be used within the Postgres code base to kind of make that possible. It's not easy, but it's possible. That's one. And then the second one is Postgres.

Startup Explore (09:42)
Mm-hmm.

Philippe Noël (09:48)
as a database stores data on the disk of the computer in a row format. So different cells of a specific row are stored next to each other on the hard drive. And this is really good for processing small amounts of data that you need to retrieve very quickly. But when you're doing analytics, oftentimes you may want to get the total value of a field. So if you imagine a database where you have sales transactions, maybe you want to get the total number of sales transaction values to get the

Startup Explore (09:52)
Mm-hmm.

Philippe Noël (10:17)
you know, the sales you've done in a month. That actually requires taking a specific column of the database and sort of summing all of the categories. And in order to make that faster, you actually want to store data, not with different blocks of the same row together on the hardware drive, but with different blocks of a column together on the hard drive. And so Parade has also sort of brought in these innovations inside of Postgres. And the detail of it is, you know, perhaps beyond, you know, beyond the point here, but that's kind of what we've done to make it better.

So ParadeDB actually doesn't change the interface. And that's part of the pitch, right? Customers are happy with the interface that Postgres offers. But we do change some of the things behind the scenes so that you get much higher performance when you were using the interface that you're used to.

Startup Explore (10:47)
Got him.

Got it. Got it. Thank you. Now, speaking about the installation and deployment process, how does typical deployment process look like for a company that might be potentially thinking to switch from this traditional setup with Elasticsearch to PrairieDB? And how long it might take, how difficult or easy that is?

Philippe Noël (11:18)
Eh.

Yeah, that's a good question. It depends on their situation. In general, what customers do today is they have a Postgres database. At least the customers we service, they have a Postgres database and they have this ETL or data pipeline between that and between their Elastic. When they deploy Parade, they essentially add us to their existing Postgres and the data is already there. So they don't need to move data. What they will need to do for the migration is they will need to...

write new code for their application that actually queries a ParadeDB database instead of an Elasticsearch. And our query syntax is slightly different than Elastic. That's really where most of the work is. Now, the reason I said it depends is because it depends how complex their Elasticsearch story is. For some companies, it can be a matter of a few hours. For some companies, it can be as much as two weeks, let's say, if they have a lot of queries.

Startup Explore (12:16)
Okay, that's still not bad. mean, if we're talking about two weeks or two to four weeks for large enterprises to install and deploy, that sounds like a quick, okay. All right. In terms of the future product features, which one, I guess, what can developers expect from ParadeDB?

Philippe Noël (12:23)
Yeah, it's not bad.

More of the same and faster. have pretty good search. Search capabilities are very good now. Analytics capabilities are decent, but they're not anywhere near as good as they should be. By the end of the year, I think they will be. So we have a lot more work we're doing on analytics. Obviously, we do a lot more work nowadays also with the vector search and the AI search capabilities, which are very relevant with the whole AI workloads that happen around the world.

Those are probably the main two ones that we're building. And then the next big thing I could say is like ease of adoption. The product was historically pretty enterprise oriented. It did require a little bit of effort to get running. We're spending a lot more time making it really easy now and integrating it in places that people expect it to be with other industry partners so that it's that simple to be able to have access to it.

Startup Explore (13:14)
Mm-hmm.

Got it. Exciting stuff. Now let's switch gears to, to your background. Can you talk about your, the school that you went to, your prior experience with starting West and ultimately what led you to start building ParadeDB?

Philippe Noël (13:47)
Sure, sure. Yeah. My background, I grew up in Quebec, French part of Canada, moved to the US for university. I studied at Harvard, I studied computer science and neuroscience. Yeah, my senior year I started in different startup, as you mentioned, called Wist. It started as a remote desktop solution for creative professionals. So the idea was it was the beginning of COVID and

The idea was people couldn't really work from their big computer workstations in the office. So we were running big workstations in the cloud with graphics cards and giving them remote access to them. This eventually turned into being a browser that was powered in a similar fashion because most people that we talked to actually powered their applications through their browser. And eventually we realized that this was quite a small market, frankly, and we made so many mistakes on the way that it really wasn't taking off. So we closed the company down.

Startup Explore (14:40)
Haha.

Philippe Noël (14:44)
And took a few months and eventually we started Parade. We started Parade, so in few months that we took off, we built a bunch of applications, basically things we were interested in. We spent a lot of time working between search engines and vector databases and Postgres, realized it was really, you know, it had a lot of complexity. And then we joined Y Combinator, talked to other people and realized this was actually a very shared problem. So we started.

decided to start working on it.

Startup Explore (15:15)
Got it. majoring in computer science, neuroscience in Harvard. That's pretty impressive. I know your co-founder Ming is also a major, in computer science from Harvard. How do you guys divide, divide the responsibilities of running the company between you?

Philippe Noël (15:27)
Yeah, that's it.

It happens pretty naturally, to be honest. I feel very fortunate to be working with him. He's a very smart guy and he's very smart at things that I'm not as good as and, you know, vice versa. so Ming is our CTO. He's really the technical mastermind. We talk a lot about product directions together, but at the end of the day, when it comes to implementation details, he very much has the final say and he always makes the right decision. So it's great. You know, it's great to have him be in charge of that.

He does a lot of our content marketing as well as a very good writer. He's a better writer than me for sure. And, you know, a variety of other things based on, you know, who excels at what. I do tend to do more of the sales, for example, I handle a different part of the tech of the product, although less and less nowadays. you know, maybe I should say like things that are internal facing he handles and things that are external facing I handle. It just kind of evolved that way over time.

Startup Explore (16:28)
Yeah, makes sense. Good complementing each other's skills. You've mentioned YC. I'm curious to hear your experience in YC. And I know they are pretty selective when it comes to taking startups to their school. So how was your experience there?

Philippe Noël (16:45)
YC is amazing, honestly. They gave us so much. I really like it. We joined YC, we had inklings of an idea of what we wanted to build, but frankly, we weren't too sure. We hadn't fully decided on this yet. YC was really helpful in bouncing ideas. We met some really amazing people that are very impressive. And the community's been great. I highly recommend it.

few downsides to doing YC. I think the only one people mention is they take a good amount of equity. honestly, think the well, at least when we started, you know, our first company had gone to nothing. So we told ourselves, you know, 7%, which is how much equity they take. 7 % of nothing is nothing, right? And most startups, most startups go to nothing. So if it increases the odds that it becomes something, it's actually a pretty good deal. And I we feel strongly that was

Pretty good way to look at it now.

Startup Explore (17:44)
That's great to hear about the experience. talking about the total addressable market for you guys, I know it's big because a lot of companies are accumulating lots of data. They need to perform certain analytics over that data. And what's more exciting is that the market is probably growing for you. I'm curious to hear your thoughts about the total addressable market for you and as well as ideal customer profile.

Philippe Noël (18:11)
Yeah, yeah. So we sell to companies that use Postgres as a database. Today, Postgres is the second most deployed relational database behind MySQL, and it's the fastest growing one. So it's probably going to become the most deployed one within a few years, if not earlier.

So the market is big. And then from that total subset of people that use Postgres, whichever subset has search analytics needs, which as you point out, is a pretty big share of it, but it's not all of them.

Exact numbers is TBD, but it's large. We typically work with companies that are, at least from a commercial standpoint, we work with companies that are, let's say, at least 50 employees and up. We found that before companies have about 500 gigabytes of data inside their database, the existing functionalities of Postgres are kind of sufficient for them to do what they need. It's not perfect, but it's fine.

Typically, we sell to the company's CTO, we sell to the company's VP of platform engineering or VP of engineering, chief architect, those kinds of roles.

Yeah, I do think the market is growing and we'll be entering more and more of the lower end of the market as we, or earlier end of the market, I should say, as we kind of do these partnerships with other folks. But yeah, that's how we think of it.

Startup Explore (19:34)
Got it. Now, I believe Alibaba was not your first customer, right? Who was your first customer? And then if you can share like some of the steps that you guys took to land that customer.

Philippe Noël (19:46)
Yeah, they actually were our second customer, so they were pretty early. So I can't share our first customer because our first customer sells to the US government. And as part of working with them, they did not want us to share because they don't want their tech stack to be public for security reasons. They would basically facilitate attacks. But our first customer was, you know, there are a couple hundred employees in an organization. They actually, we...

we actually got connected to them through an investor of ours, one of our angel investors. So perhaps an unconventional story, most of the customers we landed after that with a more traditional sales method where we identified companies that we thought were promising. So we identified companies that were in the fintech sector because we saw that this was something that resonated well with our product.

They were around a hundred employee, which we thought was a pretty good size for a combination of having the pain point we're solving, having maturity and resources to go and solve that pain point and also not being so large that they might be unwilling to work with a small startup. We thought that was kind of a pretty good sort of like, you know, good intersection of all of the, of all of the parameters of the, of the customer. And then we just emailed, you know, the CTO, the vice president of platform engineering and

And that was the end of it. And then we talked to the team and the technical evaluation and everything. But getting in front of them was actually a pretty small, pretty simple, straightforward process.

Startup Explore (21:23)
Got it. That's good to hear. Let's maybe talk a little bit more about the competition. So I know we've covered Elasticsearch at a high level. Are there any other direct or indirect competitors to you and other key highlights of why ParadeDB standouts that you want to mention?

Philippe Noël (21:43)
Yeah, there's many. So Elasticsearch is a public company. They're very large business. They've been around for 15 years. They make like over $1.3 billion in revenue. And actually, Amazon also sells Elasticsearch and they make $2 billion in revenue. So you can say the total revenue from just that product is probably like $3.5 billion or something. So it's a very large business. Obviously, as a result, there's also a lot of other competitors.

There's many, can name some if you want, but there's like five or six that are pretty high profile. The main difference with Parade is all of these databases that are competing with Elasticsearch are not integrated in Postgres. They're just like a standalone competitor that is honestly built in a very similar way to Elastic, but they try to optimize at the margins. our thesis was always, if you want to build something that's 10x better, you need to make a much different...

architecture decision. And so Parade has decided to go all in on Postgres.

We've done it because we think the TAM justifies it. And what we're building is much better than Elastic for Postgres customers because we get to make design decisions that only make sense for those customers. So for example, removing the whole synchronization layer. That's only possible because we're built inside Postgres. All these other competitors that I mentioned, they also require a synchronization layer.

Now they can sell to anyone, but we can sell to only Postgres people, but significantly better. And there are so many of them that we decide that's actually a worth a trade off. I would say that's the main differentiation.

Startup Explore (23:18)
I'm Xanthan.

Got it. That makes sense. I feel like that's a differentiating point for ParadeDB beings, very unique in that regards. In terms of the market traction, we've talked about Alibaba use case, your first customer. Are there any other key business metrics that you would like to share? I don't know. know I'm mindful that the company is still kind of early, early stages tech, but I wonder if you can share any...

business numbers right now.

Philippe Noël (23:54)
Yeah, I mean, make good money. We sell to a bunch of different companies. You can see some of the customers that we have on our website. It's not the whole list, but some of them. We sell to fintechs. We sell to e-commerce. We sell to observability, sales and elevations. We sell like five, six-figure contracts usually. Yeah, there's nothing like very specific that I think would be super insightful to share other than, yeah.

I don't know if there's anything specific to share. We've raised a couple rounds of funding. A lot of this is not fully public. Yeah, that's the high level of it.

Startup Explore (24:30)
Got it.

Got it. And in terms of the funding, are you at seed stage or further down the road?

Philippe Noël (24:36)
Where is the search company?

Startup Explore (24:38)
Got it? Got it. Okay. And looking into the future, where do you see ParadeDB, let's say in five to 10 years, and what would be the North Star for the company?

Philippe Noël (24:50)
Yeah, that's a good question. mean, five to 10 years is such a long time and startup journey that it's hard for me to say something that is likely to be true. We're following the footsteps of Elastic in a lot of ways. We're sort of going about the marketing and the product building in a very similar stages. So today we're building the foundational database and we provide this as a user facing database for customers.

There's a lot of other markets that the company like Elastic has entered into. They also do security software and they do log processing as well, which are very lucrative markets. We have our ideas for how we can go and enter into those markets as we grow as well, but they kind of require the core foundational database to be really mature before we can expand in those ways. So if everything goes well, based on what we know today, we have plans to eventually go in those directions maybe.

one, two years, three years, I don't know how long exactly it will take to build the foundational piece. And then I think there's a really huge coming to build here, like $20, $30 billion valuation company.

Yeah, but a lot of this is early hypotheses. We have a lot of work ahead of ourselves before we can earn the right to play in those fields.

Startup Explore (26:10)
Yeah, I agree. Like five years in a tech world is a very, long time timeline. It's hard to say what's going to happen. But that's a helpful perspective. Thank you for sharing. as you grow, you need to hire more and more folks and people. Are there any particular specific aspects of candidates that you pay special attention to? Anything that you would highlight?

Philippe Noël (26:17)
Yeah.

Yeah, so we're a very small team. We're only four people. And we are hiring. We hire mostly engineers today. From a technical perspective, we care about engineers that are familiar with Postgres or are familiar with database internals. There's actually quite a small group of people that fit that profile. So that makes hiring somewhat challenging. Or challenging, maybe not the right word. There's a good number of people, but slow. We have pretty high standards.

I will say the two things that we care a lot about on top of this is willingness to interact with customers directly. Not every engineer is interested in doing this, but it's something that we feel very strongly about. We like to have every engineer be at least, you know, if not interacting with customer directly, at least seeing what the customers are saying. I think that's important to building the best product and also to providing the best service. So that's probably a big one. And we like to work with nice people, you know.

Startup Explore (27:33)
That's nice to hear. And that's actually a unique approach. So you want every basically engineering person be able to do customer support and interact with the customers if need to. it. Got it. That's true.

Philippe Noël (27:35)
Yeah.

Yeah, yeah, yeah. I mean, we build a very technical product and it's used by engineers as well, right? So it's

kind of like, you know, talking to peers in a lot of ways, just at a different company. So it actually works quite well of our engineers talking to our customers engineers. It's not like an engineer may be talking to someone in a different business division, so they might not be as familiar. And also, because we built such a technical product and we have such a technical customer base, the questions are just...

Startup Explore (28:03)
Yeah, yeah.

Philippe Noël (28:12)
so precise that customers will have, right? And oftentimes you kind of need the person that was responsible for building those functionalities to really provide the best service. And that's the engineer. So if we had customer service, I think the quality of our customer service would be much lower if they were people that were not familiar with how the product is built.

Startup Explore (28:30)
Yeah, that definitely makes sense. since the product is so technical now, let me ask you this. If for potential future candidates out there now, why should they join ParadeDB? What would you tell them?

Philippe Noël (28:44)
What would I tell them? I mean, we're four people, right? It's the opportunity to have so much leverage. Everyone that works at Parade gets to do so much, so much more than they do at other companies and the design decisions compound over time. So from that standpoint, I think it's really exciting. Those are the type of people that we also try to attract, right? That's I have a big impact. So that's one. The second one is obviously, you know, the company is doing very well and...

When you join a company that's doing well at such an early stage, also set yourself up for a very big upside. So I think that's another big reason. And then the third one maybe is there's few people that make big companies out of Postgres or out of databases. So if that's an area of interest, it's probably one of the best ways to capitalize on it.

Startup Explore (29:30)
Got it. Makes sense. Now, Philip, thank you. We've talked a lot about the company product market competition. Let me ask you a question outside of work. What hobbies or activities you have and you like to do?

Philippe Noël (29:45)
I do a lot of sports. I play badminton pretty regularly. I play soccer sometimes. And I run and I exercise and I swim and all of the things like that. So that's kind of how I keep busy. And then the rest is just regular life, maybe you would say, right? Yeah.

Startup Explore (29:47)
Nice.

Got it.

And you haven't picked up Pickleball yet?

Philippe Noël (30:10)
I've played pickleball a few times. I've played pickleball a few times. I wouldn't say I've picked it up though. No, I've stuck with badminton more. Yeah.

Startup Explore (30:16)
Okay.

Got it. Got it. Got it. Okay. And then my, thank you, Philip. And last question is what advice would you give to other tech co-founders based on your experience, especially for those who maybe want to break into enterprise sales space, what would you tell them?

Philippe Noël (30:39)
I don't know if I have good advice. I'm trying to figure that out myself, to be honest. what I've learned so far at least is you want to do very, very few things, but very, very well. And I think that stands for both product and for sales. So at least in our early days, we were so tempted. You want to build something big. So you want to do a lot of things and do them. But the noise level of the world is so high.

reach above the noise level of the world to get people's attention. And that requires doing a very narrow thing, incredibly well. So it sort of spikes up, right, in their interest and in their quality. So now we try to build very few things. My co-founder actually, I attribute a lot of our success here to my co-founder, who's very, very good at doing this and sort of slowing us down when we're trying to get brought in too much. But both in sales and in product, we wanted to be incredibly focused, at least for us so far.

doing this as a spin-off.

Startup Explore (31:39)
Thank you so much, Philip. I'm really excited our time and hopefully we will meet sometime soon again and you will have even bigger successes and milestones to share.

Philippe Noël (31:49)
Yeah, hopefully. Yeah, thank you for having me.