The Ravit Show - State of Data Survey, ETL, ELT, AI with Michel Tricot, CEO & Co-Founder, Airbyte
State of Data Survey Airbyte
Ravit Jain: [00:00:00] Finally, we are you're, and I'm so happy to host you on the Ravit show. But for our audience, I'm pretty sure you're not a new face to the community. People know you, people know, but still just little background about yourself. Can you introduce yourself?
Michel Tricot: Yeah, of course. So I'm Michel Tricot. I'm the co-founder and I'm the CEO of Airbyte, I've been working on for the past three years now.
Michel Tricot: In reality, it's a little bit less than that because we had a few people at the beginning, but that's that's where we're at. And yeah, I've been, I'm originally from from France. I moved in the US in 2011 and I've been there since then and always been in the data space, different scale, finance scale, internet scale, nice IOT scale. And yeah, that's me in a nutshell.
Ravit Jain: Okay. Yeah, I think it's pretty cool and what your experiences have [00:01:00] been is what we see as well with Airbyte. Like in last three years, we've seen the growth for Airbyte. And I've been closely following it, not, and that's what, that's how I, said this was long due since, because it was since the initial stage I've known you and John.
Ravit Jain: So it's pretty cool to see how Airbyte has a nonstop growth there and still going very quick in the market and making those strides. So pretty good stuff that Airbyte and team is doing. Also, Michel, I wanted to, definitely I'll be asking more about the state of data survey that you recently came up with, but Before that, I will take a step back and ask about how did you come up with the idea of Airbyte?
Ravit Jain: Like where did that idea generation happen? I'm pretty sure the audience would love to learn a little about that as well.
Michel Tricot: [00:02:00] Yeah, I think I, I grew a little bit tired of rebuilding the same system over and over again over the past since I started my career in in 27. Like in, yeah, in 2007.
Michel Tricot: Every time I would go. Start a new data project, and it was always the same thing. You build a connector, you put your data somewhere, the connector breaks, you need to fix it, you monitor it, et cetera, et cetera. And then you add more and more connectors. And I did it. Yeah. I was at a company called FactSet, very big on like financial data.
Michel Tricot: Yeah. Same problem. I was at LiveRamp, same problem I was at RideOS same problem. And at some point, yeah, that's when we, John and I, we just we've been. Working, not working together, but we had a few site projects together starting 2013 and nice. 2019. That's when we say, okay, both of us have the same type of forum.
Michel Tricot: We always struggle with bringing data into systems and let's [00:03:00] let's fix it. One sense for all for everyone.
Ravit Jain: Oh my. That's amazing. It's yeah, it's pretty interesting, obviously, and it wouldn't happen if you wouldn't have been, facing those problems day in and out.
Ravit Jain: You, you'll have faced it you and John, and that's why you have found that common ground, that this is it. Now it's time to, take a step in fix. And you are liked fixed it for so many companies are there. So yeah, we'll definitely come on the numbers as well about the growth that I might have seen.
Ravit Jain: But to start with we recently saw the state of data survey. I was happy to provide a quote for it. So I, I feel honored by that. I could do that because it's one of the best, one of the best reports that come out almost every year that y'all do. So the positive sentiments towards, Airbyte what I've seen is pretty impressive.
Ravit Jain: But at the same time [00:04:00] do you see a lot of companies moving from other ELT solutions? What's your take there? I would love to learn a little about that.
Michel Tricot: Yeah. So I think we got about 900 people on the survey. We tried to, yeah. Of course we were the one doing it, but we tried to be very unbiased and just putting the raw data, not trying to just go for our own community, but other community.
Michel Tricot: So we had a good sense of what's happening. The other thing is there are people yes. That are moving from close source solution. I think right now we are at a point where people are, Sitting in like in between multiple chairs, like they're testing multiple solution in parallel. And whether it's close source, whether it's open source or whether it's just.
Michel Tricot: Internally built. [00:05:00] And I think that's the, that's what we see today. Now with regard to Airbyte, our goal is really replacing what people are building internally and bit by bit grabbing what comes from the closer side. And I think, yeah, when you see that people are just, I think we're what, 6,000 deployments every single month.
Michel Tricot: And it keeps growing every single month. So there is a trend that is people by default. They try to build something and with Airbyte they don't have to build it. And that's the default behavior for data engineers. And I think right at that point, that's what is just pushing the adoption of like open source, ELT solution.
Michel Tricot: Okay. And yeah. Yeah. And there is there is so little friction. An engineer can just download the code deploy it, and they don't have, they don't need to ask for permission. They just, it saves them time. They don't have to be on the Salesforce connector. Everybody happy.
Ravit Jain: Okay. Pretty interesting. And thanks for [00:06:00] sharing those insights.
Ravit Jain: And I'm sure first of all, the report I like how you all, go out and keep that even if it's coming from air nearby, but still, you all have a very vendor neutral ground. I've seen the report and that kind of, brings up a lot of interest for the others as well in the space, in the data community.
Ravit Jain: They kinda feel that, oh, they can look up to something which is even produced by Airbyte, but still there's so much info about each and every vendor about the communities they can follow, what's growing, what's happening and it's like a good refresher for every data engineer to look at the things that they could do.
Ravit Jain: I saw even podcast being there newsletters and everything was mentioned there, but. Coming back to something, what you were talking about which is around the ELT solutions. Can you share any of the experiences, like any of the common challenges that [00:07:00] you've seen companies face transitioning from other ELT solutions to Airbyte and how do you address.
Ravit Jain: Those types of challenges, because I'm pretty sure you might be getting there might be so many companies, enterprise leaders, data team, coming back to you and saying that, oh, these are the problems that we face. How do we solve that? So how do you do that, Michel? Yeah,
Michel Tricot: so there is obviously the first initial one, which is when you're using a solution in general, that solution has some opinion about what the data looks like.
Michel Tricot: If you look for example, at Fivetran, you. They have an opinion about what the data should look like, how it should translate into table, into your warehouse. And after that, people just build their data pipelines and their data processing on top of this particular schema. So that is always something that we that we see with our customers and with [00:08:00] our open source user is how do they bring Airbyte and how do they create that compatibility with the.
Michel Tricot: Replace an existing system. So this is a, I would say this is a challenge actually. Recently we had a hack project from from Airbyte and one of the engineer just was experimenting with, can we do some automated translation between what Airbyte generates and what, for example, a Fivetran ERD will look like and how close can we get to that?
Michel Tricot: So to ease the migration, but this will be the, I would say the one of the pain point that people might have, but it's solvable. And that's also why Yeah, we're working a lot with DBT so that we can very quickly developed like migration models if necessary. The other one, and that comes more in larger companies when [00:09:00] they move not from the modern data side, but more from legacy ETL and ELT solution is that they were buying end-to-end solution.
Michel Tricot: So something that was doing, yeah, everything from ingestion up to dashboarding and then they say, oh yes, I want the modern data stack, I want to get my I just bought my new fancy snowflake. But now I suddenly have to deal with five different vendors. And that's generally a bit of a problem for them because when they see Airbyte or when they see Fivetran, okay, but I also want to transform my data.
Michel Tricot: How, what should I do? And then they need to talk with with DBTs, they need to. To find another vendor to do the transformation piece. So I would say that's a challenge for some companies, but more for enterprise level or high midmarket companies, which is you're replacing one solution by multiple smaller ones.
Ravit Jain: Yeah, no, I think that, that [00:10:00] makes sense in terms of, obviously those big enterprises would have so much in the background that they might be working around with. And if there's something, which is where they're wanting that change to happen, it'll take time and. Obviously, Airbyte kinda sits in there, obviously are closely working with DBT to make that happen as well.
Ravit Jain: It, it does make all sense. Thanks for those points Michel that brings me,
Michel Tricot: yeah, maybe just one, one last thing is Yeah, sure. Some people move from cloud self cloud hosted or SaaS solution Airbyte is open source, meaning that there is a component where you need to suddenly operate a new system.
Michel Tricot: Exactly. And, but they generally make that choice because one it's cheaper to like to operate than a cloud solution because you have more control over your cost, especially as your data is growing. And the second one is that gives you a [00:11:00] lot more security around your data and that removes a lot of red tape for using your platform.
Michel Tricot: So that's both a challenge and something that people really enjoy with with Airbyte.
Ravit Jain: No, I think definitely all makes sense there. Another quick question, something which is related to it is how do we see any ELT platform, addressing the long tail of connectors?
Ravit Jain: What's your take on that, Michel? Yeah,
Michel Tricot: So that's funny because I actually, I worked enough article very early in the life of Airbyte and everything with regard to connector. You need to see that as. It's a sequence of layer and when we build Airbyte, we build it around a protocol, which is the lowest possible level you can think of from moving data from point A to point B.
Michel Tricot: And then what you do is you just build abstraction on top of it to address families of connector. And today we've got to a point where [00:12:00] people can build connector with just a ui. And so there's the no-code connector builder. I think we have about, we released it just two or three weeks ago, and we have about.
Michel Tricot: 700 like live pipelines with just connectors that were built. So that's pretty nice. But what people need is the simplicity of adding their own connector and maintaining these connectors. So the more abstraction we can build on top of that protocol. The more we will be able to address the long tail, the more we'll be able to empower the community to share like this little snippet of YAML that allow them to just extract data from a source.
Michel Tricot: So that's I think that's the only way a ELT, ETL platform can actually solve the data movement. Problem is if you provide the ability to people to just customize, edit. And you give them the urgency to just move and change their connectors. One, one thing we're doing [00:13:00] on our side, we're also, it's something we started recently, which is every single customer that comes to Airbyte, if they ask for one, one connector that we don't have.
Michel Tricot: We build it. Wow. It's like basically this on demand connector. And, but we can do it because we've built the abstraction. If it was like building Python code to do that connector, we would not do that. Now that we build this abstraction, we can do it and it take us, what, 15 minutes because the people know how to use it, no code and know how to read the documentation and they can do it.
Michel Tricot: So that's a pretty big deal for us. And after that, it's our community. How do we crowdsource the development of these connectors?
Ravit Jain: Oh wow. Yeah, that's one of the, also the benefits of having that open source community where you can learn and you can build something together, but at the same time, this is a very interesting feature that I see, obviously on demand.
Ravit Jain: Like I haven't heard any of the companies ever doing that building it's almost like a product that you're building for [00:14:00] something that a customer needs or if there is a requirement. And maybe that's one of the reasons why we see the scale for y'all as well at Airbyte where y'all have so many connectors, but at the same time there's so many companies using it.
Ravit Jain: So what's the number? And okay. I would love to, obviously for our audience, I know it like the scenes that followed y'all, but from the start of your journey from 2019 if we go by a year by year, What's like the if you like the numbers in your obviously if you remember the numbers, Michel, we would love to learn about how many companies started using it in the first year.
Ravit Jain: Second year. In the third year. Yeah.
Michel Tricot: So let's, I think we need to yeah. Okay. I can do that. We released Airbyte. End of, I'm going back into my, my, my memories on numbers. So we
Ravit Jain: [00:15:00] released spot. Yeah.
Michel Tricot: We released Airbyte in October, 2020. And Right. Pretty quickly we started to get what, 10, 20, 30, 40, 50 users of Airbyte.
Michel Tricot: And they were just deploying Airbyte. They were just experimenting with a few connectors. And to be clear, when we raised Airbyte, it was. Very alpha. We wanted to get the community insight from there in 2021, so we began 2021 with maybe I don't think we can even talk about daily activities or these were more like people experimenting.
Michel Tricot: So I'm just gonna look at end of 2021. Wow. Beginning of 2022, we had maybe. 700 daily active user, 500 daily active user meaning company that have taken Airbyte and make it part of their stack. And keep in mind that end of 2021, the product was still in [00:16:00] the very early phase. Early stage, yeah. End of 2022, we had about 2,500 daily active user.
Michel Tricot: And today we have about 3500 daily active users. Oh my. So it just continues to grow and it grows really fast. And, but that also makes sense. Like what you have people who wanted to try Airbyte in 2021, who was not solving their problem because Yes the platform was not as much mature as they expected, and now it becomes more and more mature.
Michel Tricot: So they come back, mature, and boom, they start adopting. So yeah, that's, that would be approximately the numbers.
Michel Tricot: We are about to hit 100,000 Deployments of Airbyte, oh wow. So deployment is just someone downloads Airbyte and runs it once, so that's what source.
Ravit Jain: No, I think first of [00:17:00] all, congrats in building something, which is and I, that's one of the reasons I mentioned I've seen the growth and growth is like, this is hyper growth, to be honest. And for all the good reasons obviously all I am pretty sure one reason is also, like you mentioned in the previous answer, is about creating a connector on demand, which is where I'm pretty sure if there's.
Ravit Jain: A company that comes to you, or there's a contributor that comes to you and asks for a connector. I'm pretty sure there might be four companies who would also want to use that connector and now they've found it on Airbyte. So that kinda takes off very quickly. So well done on that. Michel, definitely very interesting and thanks for sharing those numbers.
Ravit Jain: But again, that kind of also puts me kinda makes me curious as well. And something related to that is about the Airbyte 1.0, which is not yet [00:18:00] out. What's next? When is it happening? I'm pretty sure the community also wants to learn about that can you tell us? Yeah.
Michel Tricot: So 1.0 for us is really a product and usage milestone.
Michel Tricot: We we know what we want to do for 1.0. And right now what we've been doing over the past two and a half years was really focusing on building the fundamental features of a data integration platform. Something as simple as column selection, something as hard as schema migration and schema evolution.
Michel Tricot: It's also about speed, like performances. We've, like that's something we're gonna talk about it, we're gonna talk about in the next few weeks, but we've done a massive revamp of. The performances of Airbyte for replicating data from databases to to [00:19:00] warehouses, that's gonna beat every single system on the market.
Michel Tricot: So stay tuned for that. But Nice. Yeah and so these are, for me, these are fundamentals. But then you also have, and this I'm talking for open source especially, is the ease of deployment. The ease of operation is also important, and we want to have that for 1.0. So we have something that. Works today.
Michel Tricot: Yeah. But it still require people to just hack around a little bit, to just, how do you monitor that within your monitoring tool, et cetera, et cetera. And we wanna make sure that there is a sense of, you download Airbyte it can run on your Kubernetes cluster. It connects to your services. And it just works here.
Michel Tricot: It still require you to just dive a little bit into configuration files, et cetera, et cetera. So that's what we want to get to for 1.0 is it's a fully productionazable product. So today it is, but it requires people to, yeah. [00:20:00] To work around some like how do you connect to a vault to protect your keys?
Michel Tricot: We want that to be part of a 1.0 because yeah, security of your data is key. Nice. So that's the kinda thing we wanna be building before we do 1.0. Now, cloud, although it's based on like a version that is not 1.0 of open source is general availability because we've built all of that internally for cloud.
Michel Tricot: So now it's just how do we bring that and make it available on the open source distribution as well.
Ravit Jain: Okay. Yeah, I think these are pretty interesting points that you've shared and definitely makes sense to have a 1.0 because. It will be a blast for at least the folks who are using Airbyte. There are so many capabilities that they will have.
Ravit Jain: So definitely looking forward to that. Thanks for sharing it first on the Ravit Show. I'm happy the [00:21:00] secret sauce. Now people get the secret sauce and also get something to learn more about which what they can expect in one point away, right? So definitely looking forward to that.
Ravit Jain: But. That. Another question around like shifting gears a video in terms of talking a little about the data security and privacy as well. So how does Airbyte ensure data security and privacy for its users? And at the same time, I'm pretty sure if folks would love to learn if if there are any specific measures or certifications that are in place that you can share.
Michel Tricot: Michel. Yeah, so for this, we need to look at Airbyte on from two different point of view. One which is open source, and also our self-managed offer and cloud because both are very different for open source and self-manage. We build Airbyte in a [00:22:00] secure way, but now the, there's, because we don't control the environment, which it runs.
Michel Tricot: It becomes the responsibility of the person who is deploying Airbyte to run Airbyte in a secure environment. But this is actually why people are using Airbyte Open source is cause they trust their environment so much around security that they know they can run Airbyte safely within their infrastructure and they have all the security check in place to make sure that Airbyte open source runs and it's secure for cloud
Michel Tricot: we obviously have a very secure architecture we're building around data. And data is, has to be secure. So this is where, we have very specific roles that we've hired for one for with regard to privacy, especially as people are moving data within Europe or within within the US we want to make sure that privacy wise we're doing everything to protect our customer.
Michel Tricot: And we also have one person just [00:23:00] dedicated on security like, How is cloud secure? Are we rotating keys? Are we protecting our voice? Are we monitoring our votes? Are we monitoring access? So all of that is there. And yeah, we very early on in the life cycle of the, of Airbyte Cloud, we went for like SOC 2 compliance, ISO compliance, et cetera, et cetera.
Michel Tricot: So we have checks. And in place internally to make sure that whoever is using cloud can be confident that yes, we're secure and that we're protecting privacy. And also we never stop data. So that makes things a little bit simpler for us.
Ravit Jain: Yeah. No, I think this is very important information and one of the, one of the things that.
Ravit Jain: Obviously we've been listening in is around data governance, data security, so you all have all the measures in place, which is very cool. So thanks for sharing that, Michel. Also another question around, since we've been. Obviously you've [00:24:00] spoken about the first three years, the early, the early three years.
Ravit Jain: And to be honest, I would say it's two years where you all actually put things together towards the end of 2021. But how has it been you've shared till now, but how does What's your mission in the next three years? I would love to learn a little about that. For Airbyte what does the mission look like?
Ravit Jain: But you mentioned about 1.0 definitely. But, for the next few years, how do you look at Airbyte growing?
Michel Tricot: Yeah, so 1.0 is about getting very strong security fundamentals and feature for data integration. I think what's gonna happen in the next three years for us is gonna be more around the integration downstream.
Michel Tricot: So how do we provide more visibility on where the data is [00:25:00] going, where it's coming from. So everything was regard to data in age. Wow. We've also, we've been focused a lot on data ingestion, which is one the left side of what you do with data. We also want to look at what can we do on the right side, which is how do you activate that data?
Michel Tricot: So we don't want to do dashboarding or anything like that, or like BI, but everything that is related to, for example, what we call Reverse ETL. Which is the ability to pull data from warehouses and sending into services. This is also something we want to look at because the problem that we need to solve is very similar, which is there is a huge amount of places where you can push your data and the more you're able to push it, the better your company is gonna become.
Michel Tricot: So we want to see how much can we apply what we've learned on the ingestion side and how can we bring it into the pushing side. So this is a big deal for us. And this way we basically create all the pipes for the data to move from point A to point B, and then [00:26:00] we integrate with all the other application that are building the intelligence on top of it.
Michel Tricot: Yeah the other piece is obviously, it's. What do we do around AI? Because AI is probably the biggest consumer of data. So how do we bring data, not just for analytics purposes or operational purposes, but for actually for training algorithm, creating knowledge base for AI algorithm. So that's a, that's also a big deal for us in like in the next few years.
Michel Tricot: Yeah.
Ravit Jain: Yeah, and I think rightly mentioned, obviously first of all data integration, one of the most important things out there. And I see what your focus is. Definitely looking forward to see how it evolved, but just a quick question on the data integration bit as well.
Ravit Jain: We've seen the landscape evolving like way very much a and rapidly, to be [00:27:00] honest. So what are some emerging trends or challenges that you foresee in the industry? And yeah, you did mention a little about how everybody is, obviously preparing to address them, but. Are there anything that, is there anything that you would like to share?
Ravit Jain: Yeah,
Michel Tricot: so I hope I'm not stating the obvious here, but I think iceberg Delta, like all the data lakes technologies are going to become a lot more like going to become the default of where you push your data. You. You can still push data into data warehouses, but warehouses are gonna push a lot more for being compatible with these files because this is like the unique language for talking to data, like SQL is one language, but having your data that is [00:28:00] stored in a way that is queryable, that is efficient, this is the way to go.
Michel Tricot: I think that's what we're gonna see in the next three years. Now who wins between Iceberg and Delta. The future will tell us, I don't know. I think right now we need to work with both, but that will become the default for what you push on data. The thing that I believe is going to become a lot more We're going to be a lot better at managing unstructured data.
Michel Tricot: Everything that is text data. And that will be that will be powered by Like that would be powered by AI. Like how do you take, how do you extract information from an English text and how do you structure that data so that your analyst can actually do something with that, with this insight? And I wonder what kind of companies are going to be built on top of [00:29:00] Data Lakes data warehouse to actually, yeah, not just transform the data, but structure the data so that it's.
Michel Tricot: Can be, you can do automation on top of it. So that's gonna be a, I think a big a big deal in the data space for the next few years. But it's only enable, because we developed the technology to to extract data from text in a way that text makes
Ravit Jain: sense to people. No, I think definitely good points there, Michel.
Ravit Jain: Thanks for sharing that. Also when talking about a lot of data, unstructured data, structured data, that kinda brings me to another, and you've spoken a bit about it, but another question which is a very common question nowadays, which is around AI moment. So how do you, how do like, How do we approach this new AI moment?
Ravit Jain: What's your thought on it and where do you see it in the next one to two years? I'll just keep it [00:30:00] because with AI, things are evolving so quickly. Every company wants to actually, adapt AI. I saw a survey recently. It said 75% of the companies now in 2023 won't have
Ravit Jain: some AI featured in their product to offer to their customers. So that, and I've been, hearing a lot about that, but how do we approach it? What's your take on that, Michel?
Michel Tricot: Yeah. I think AI is gonna go through something very similar to what happened to data over the past five years, which is, it's gonna be a very heterogeneous environment where you will have, I don't know, 10 different ways to do vector stores.
Michel Tricot: You will have 100 different ways to do embedding and the same way we're talking about the modern data [00:31:00] stack today. I think there will be the modern, I don't know if we can even call it modern, but like what is the AI stack and what, how will that breakdown between. Like data ingestionpiece, data activationpiece, data storage, ah, and embedding ALM like even the workflow managers how will all the things that we've solved today for data are going to apply or are going to have some flavor of it in the AI world?
Michel Tricot: And I think the next three years are going to be keep your people should keep their option open because we, who's going to be the players that you want to be working with. And everyone will have a small differentiator, but I think the market for every one of this category is going to consolidate.
Michel Tricot: And then you want to make sure that you are working with the best player over there. But right now, keep your option open. Tried the best one today, but make sure that you can evolve in the future. Don't [00:32:00] settle on what your stack should look like. Cause it's gonna evolve just the next
Ravit Jain: Oh wow. I like I like that perspective like, Now since you've mentioned now, I see that, oh, you're right about, something which could be like the modern data stack we are seeing.
Ravit Jain: It could be the AI stack or it could be something else which would be created. And obviously there might be many more categories which might come in and how it'll picture and how how data companies would also be transformed with another player in there. It's interesting. We are in this interesting times for sure.
Ravit Jain: So can't wait. But thanks for sharing that point. It I, many of the times what happens is a lot of leaders who come on the show, they share such interesting points in. You after six months, a year, or, after 18 months, we see that happening for real. And that's when I'm, I go back and I'm like, oh, I did, I hear this.
Ravit Jain: Was it a deja vu or something? But things [00:33:00] manifest very quickly. But thanks for sharing that, Michel. Quickly getting onto something which is around We've spoken at length, but I'm pretty sure audience would also love to learn a little more about Airbyte.
Ravit Jain: The are there any resources that they can hop on and, learn more about Airbyte Connectors ELT? Yeah,
Michel Tricot: The first thing is obviously the, our GitHub report. It's okay. I think it's airbytehq/airbyte. That's a big one. An education hub on Airbyte. So we try to publish a lot of high quality tutorial, high quality thoughts, pieces around data.
Michel Tricot: So even if people are very [00:34:00] beginner in the data space, They can learn a ton on Airbyte. Like we try to put as much resources as possible. And yeah, we also, yeah, and we just recently created one for like how do you use Airbyte in conjunction with Dagster, in conjunction with Link Chain? So how do you actually create the first version of an AI stack?
Michel Tricot: And yeah, that would be probably the place I would start with the repo and our content hub.
Ravit Jain: Alright, that's that's pretty cool. I'll, what I'll do is obviously when we push this slide we'll be, actually I'll share all the links with the audience in the comment section so they can actually go and learn more about it.
Ravit Jain: One last question, Michel for you is if people wanna reach out to you where can they reach out? Where can they learn more about what's new, what's coming up? Because you, you are an interesting [00:35:00] personality. I Great data professional out there I'm pretty sure a lot of folks would wanna reach out to you.
Ravit Jain: So what's the best way to get to you?
Michel Tricot: Yeah, so one thing that you'll, since the beginning of Airbyte is to stay very close to our community. So basically we have a Slack channel sorry, a Slack workspace. Community open to everyone, so it's available on our website. You can just click on it and from there you can talk to me directly.
Michel Tricot: So I don't think you can do you can do anything simpler than that.
Ravit Jain: That's the best. So reach out to Michel on the Slack channel and you get to talk and learn more and more about Airbyte, but this was amazing, Michel. It was such a pleasure hosting you today on the Ravit Show.
Ravit Jain: I can't wait to have a 2.0 session where we and this definitely, we'll talk about it more when 1.0 Airbyte is released in, we are in that [00:36:00] interesting space as well when there are, there might be many more. I would say there, there might be many more innovations that might have happened in the AI space, but not only just that, also the data integration space might have evolved very much we'll definitely have another session that time. But this was fun. Thanks for coming on the show again. It was such a pleasure to.
Michel Tricot: Yeah. Thank you very much for having me.
Ravit Jain: Awesome. And thank you everyone for joining us today. It was such a pleasure.