Greg Michaelson: Good morning, welcome back to day to day with me, Greg Michelson. I'm here with Gorav Chain, new friend of mine. And we're gonna talk this today about some really interesting stuff. So Gorav, thanks for being on, I'm happy to meet you.
Gaurav Jain: Absolutely great. Thanks for inviting me. It is amazing to be on data day with.
Greg Michaelson: With me. ⁓ Awesome. All right, well, hey, why don't we start out with an intro? Tell us who you are and what you're into and give me your background a little bit.
Gaurav Jain: So, ⁓ hey, I'm Gaurav. I am the CTO and co-founder at trio.dev. ⁓ In the industry for almost 18 years now, a couple of startups before this. At trio.dev, we are essentially solving a pretty interesting problem around how do you target ⁓ dev GTM. So, you can think of us as... ⁓ revenue intelligence platform trying to help companies that are selling to any kind of a technical audience, it a CTO or a CISO or ⁓ a VP engineering or architect or you know any of those kind of technical folks and I mean the problem which we are trying to solve is very simple reaching out to developers selling them has always been hard and don't come onto the calls. They'll block your ads. They'll never respond to your emails. So how do you basically make a sale? So that's what we are trying to do.
Greg Michaelson: Got it. Wow. That does sound like a hard problem. Certainly the amount of spam that's increased, you get on LinkedIn and over email has kind of gone through the roof.
Gaurav Jain: Absolutely, absolutely. fact, that is essentially the problem if you think it from a developer's standpoint how do we make sure that the selling is not annoying but it is more helpful for them. That is the developer side of this story.
Greg Michaelson: Okay, well, how did you get the idea? How did you get into that? You said you've been in the industry for 18 years? Talk more about your background.
Gaurav Jain: Yeah. Right. So, see, I have been, you know, all my life a technical person. I started my career with Amazon, one of the first few engineers to work on, work in Amazon India. In fact, one of the first few engineers here. Then very quickly jumped into a startup ecosystem. couple of startups before this one of the startup which I was doing in 2009. I tried it for two years didn't work out around 2016 or so is when I did another startup which is on basically we are trying to build a low code platform for digital lending so it was completely into fintech
Greg Michaelson: you Thank
Gaurav Jain: space did it for like six seven years sold it to multiple banks across india southeast asia middle east got an exit around 2022 took a break for a like almost seven eight nine months and then i was just trying to do something new this is when me my other two co-founders we sat together So what happened is one of my other co-founder who is currently the CEO at Trio, he in his previous startup, he was selling to developers. And one of the insight which he came up with was like, know, even though they had calls with CTOs or they were trying to build a pipeline at the end of the day, he was never clear of what my pipeline looks like because... you know, somebody who came and filled the form or had a good call with the sales people, nobody from the developer is actually trying out the product. Then after like, you know, maybe 15, 20 or 30 days when they are doing the call again, they realized that developers didn't like the product or they liked it a lot and they have made the decisions already. Right. So, so the sales team was not at all aware of what is really happening. And that's what the biggest problem they were facing when they were selling to developers. And when when Acyntia told me that this is the problem and then it became very easy for me to understand it because like I was saying, like 17, 18 years in the industry, I have purchased maybe 100, 150 plus tools. and hardly ever I would have gone on a demo call and made a decision because of the sales pitch. It was always based on what my developers are telling to me, how was their evaluation. So yeah, I think that's how we thought, okay, maybe this is an interesting problem to look at. We went ahead and spoke to like 100 plus. prospects, validated the idea, we understood, okay, this is a big problem for everyone. So that's how we started getting into it.
Greg Michaelson: So is it more of like a funnel management tool or is it more like how to break through spam or is it like ⁓ an outbound ⁓ like marketing tool or what does it do actually?
Gaurav Jain: So, think of it like ⁓ a three part, ⁓ you know, solution. So, one, we will help you figure out the data which is invisible, which is for example, you know, who are the people who are active on your GitHub repository, what kind of companies they belong to, who are the people who are. making some interactions with your Docker or your code command, package installers, or anything like that. Who are spending time on your docs? What kind of pages they are spending time on? So we try to bring in the data ⁓ of the developers plus the companies they are in. So that is one part. The second part is the Intel. part. So what we do is we build a developer qualified lead kind of a thing. I mean see you should I mean one of the thing which you always talk about is developers are not lead. developers are always intent. You should not reach out to developers the moment they come onto your website and fill a form. You should look at how much intensity, how many developers from the same company are coming, what is the time frame they are coming in. So what we have built on top of this data layer is an Intel layer where we are trying to build multiple scores, trying to bring in ⁓ the intent and the qualification part on this. And the third part is where we are giving you some sort of ⁓ automation or a workflow where you can actually define, okay, this happens, then this should happen, this is how we should reach out, what sort of messages we should reach out to them, when should we reach out to them. So, yeah, this is the solution.
Greg Michaelson: Got it. Where does the data come from?
Gaurav Jain: So we are collecting data from more than 30, 35 sources right now, starting from GitHub, Slack, Stack Overflow, Reddit, all those communities open data. We have beacons, which can be installed on your documentation. ⁓ get the first party data on your product to get the first party data. And then we also try to get data from package managers, be it your Docker, be it NPM, ⁓ PIP, all those sort of package managers. ⁓ In fact, what we have also started doing now, specifically ⁓ in this whole new world where developers are delegating their discoveries and evaluation to agents. We are also getting data from ⁓ MCPs ⁓ and there we are actually getting the intent in much, ⁓ I would say, a much friendly manner where people are writing that, okay, this is what we want to do. We are getting that. And so we are getting all the data from these places and then combining to give you.
Greg Michaelson: Right? Okay. Well, that's interesting. So, you mentioned that you shouldn't just reach out to developers the first time they fill out a form on your website. Talk to me about the logic that you guys have figured out around when to reach out to the folks that are signing up for a new product, So I sign up for a ton of products and I always get instant emails.
Gaurav Jain: Right, right. So see, I mean, there's obviously not ⁓ one formula which fits for all. But ⁓ I think there are multiple things which comes into the place, one is how many developers are coming and signing up on your product. I mean, for example, ⁓ somebody from Netflix has just come and signed up on your product. It could be just a hobby project for that person. I mean, if you're just reaching out to that person, it does not mean, fact, most of the time they'll get pissed off. Why are you reaching out to them? I'm not asking for any help. But let's say if, you know, six, seven or eight developers signed up in like two weeks or they are spending a lot of time on... docs, right, or specifically your docs which are talking about enterprise, production, deployment, or, know, security-related stuff. things like that plus if they have you know a lot of telemetry signals which are coming in. So, the moment you start triangulating all of these is when you see there is a good intent which is getting developed over there and that probably is the time when you should reach out you should not do it like the moment somebody just signed up. that is what we keep on trying.
Greg Michaelson: Got it. And do you guys build like a funnel for users? Like somebody signed up, somebody did action X, somebody did action X three times, that sort of thing. How do you develop that funnel? It's going to be different for every project.
Gaurav Jain: Right. So, funnel is very different from a typical MQL SQL sort of, ⁓ you know what happens in a B2B sort. So, what we say is a developer funnel looks like, you know, a four or five stage ⁓ process. First is... where a developer is discovering you, right? This is where they are just trying to know what you do, what kind of problems you are solving, whether it will be something good for me or not. The moment they start doing that evaluation, which is they are trying to figure out all the red flags, whether this works with my stack, does it work on the environment which I am working on. Do they have APIs or whatever the integration mechanism which I am comfortable with? So that's what we call as an evaluation step. So that is the second part. Once they have done the evaluation or they have figured out all the red flags and they know that, it's good to go from for us right now, they do what we call as a build a POC sort of a thing. So that's the third stage. No developer is ever going to. buy a product unless they have built something on top of your product. So a build a POC is a very, important stage there. That's where they are trying to ⁓ do some sort of package manager installation or they are spending a lot of time on your documentation. There are telemetry signals which are coming in. They are coming onto your Slack asking some questions which are giving you a hint, okay, they are evaluating or they are building something on top of my product. Once they have... done some POC, that's when it moves to what we call the fourth stage, which is a deployed stage. And this deployed stage is essentially where they would have probably taken a small plan and they are just using it for one particular use case. Two, three developers, they have deployed it, they are using it. And once they are confident is when they move to a scale stage. So yeah, this is a five stage funnel, we say discover, evaluate, build, deploy and scale.
Greg Michaelson: Got it. ⁓ so your functionality is mainly to sort of track users through those stages.
Gaurav Jain: Yeah, so what we do is essentially you get to know how many accounts are in which kind of a stage, right? When are they moving from one stage to another? What can you do to move a particular account from let us say evaluate to build a POC or from a build a POC to a deployed stage, right? And yeah, I think that is essentially what we do.
Greg Michaelson: How do you get to the what you can do kind of thing? Are you analyzing that data or is it a large language model query or how are you getting from you're in this stage and not that stage to here are the seven things you need to do to move someone from stage three to stage four.
Gaurav Jain: Right. two, three things which we do. One is, what we present is a timeline, activity timeline view, which essentially tells you, ⁓ the last three months, these many developers have done this, they've spent this much time on this kind of a documentation, then done this, this, this. So once you have all the data or all the activities which has happened from a particular account, ⁓ We also obviously get to know the location from where this is happening, the team they are working in. So then we can actually start recommending this, that, OK, now is the time when you should go and reach out to ⁓ these people. So we have our own models where you can just go and inquire, or you can research about an account and ⁓ shoot an email or. maybe reach out to them on LinkedIn with the problem they are trying to solve and what solution you can.
Greg Michaelson: But how do you know the optimum timing?
Gaurav Jain: So, see there are multiple ways and it depends upon you know the kind of tool also which is in the consideration right. For example, if it is database I mean you typically wait for a longer time I mean it does not happen so quickly, but when it is a simple API tool what we suggest is that for example, if somebody is visiting your pricing page after doing you know some that is a five six or seven types of activities. Maybe this is the time for you to reach out right or if somebody is spending a lot of time on let us say production deployment sort of a page. So, there are like you know multiple signals which essentially calls for an action right and once you get that signal you essentially take an action.
Greg Michaelson: But how do you know? mean, right. But every product is going to be different and all of the different activation points are are there.
Gaurav Jain: So we let the user define. Yeah, we let our server define. that's where the third layer comes in, right? So we have ⁓ all the granular intents or all the signals which are there, and then you can define the automations on top of it, right? So you can define things like, okay, five developers, pricing page with it, do this, right? Something of that sort.
Greg Michaelson: So the user has to divide this.
Gaurav Jain: Yeah, user can define. We do some recommendations, but at the end, like I said, a lot of ⁓ nuanced signals are there because there are a lot of these complex products which are there. So we let the user define those actions which they want.
Greg Michaelson: What if they don't know? Let's say this green field, I've got data of how users interact with my product on my website, and I'm interested in figuring out, you know, how should I market to those users to help them have the most success with my product possible? What kind of data elements are you looking at to identify those intervention points?
Gaurav Jain: Right, so ⁓ see I mean if you think about it there are like 100 plus signals right and what we have done is for each of these signals we have provided some sort of a weightage. For example, if you are a let's say a DevOps you know let's say Kubernetes cost optimization sort of a company right. So, what we have done is we have divided these products in these categories which could be either infra company or an API company or a AILM sort of a company like. And for each of these what we have done is we have built a layer of abstraction on top of it which is ⁓ if it is an optimization company look for these kind of signals, if it is an AI sort of a company look for these signals. So, what we do is we give them some assistance, but like I said at the end ⁓ every company has their own way to you know. build out these models. What we also provide is out of the other than the product, we also give ⁓ support from our forward deployment engineers who can actually come help them configure these automation workflows and build out good pipelines and all that. But yeah, at the end, mean, it's more of a consultative to figure out what works and what doesn't.
Greg Michaelson: Got it. That's kind of unsatisfying to me. I would love it if there was like ⁓ a data focused way to do that. Or do you have an agent built into ⁓ your product that can kind of dig through the data?
Gaurav Jain: ⁓ Yeah. So I think this is something which we have just launched a couple of months back. ⁓ We are trying to ⁓ build on top of it what we are calling it as, you so we have like different open data signals or real data signals as well. But we have not reached to a stage where we can go ahead and take it completely in our hand and start running the whole workflow by ourselves. And that's something which we want to do. like I said, there are a lot of nuances. mean, it becomes a little difficult to rather take care of all the nuances and give that human touch whenever the reach out is happening or outbound is
Greg Michaelson: So let's take Zerve for example. So my company is called Zerve. We're a development platform for researchers and analysts, people that are working with data. We have a coding agent baked into the product. And one of the things we see is that one of the things that we're working on right now is trying to help users find the different parts of the platform that they will take value from. There might be some users that
Gaurav Jain: you
Greg Michaelson: you know, come in and they ask a question to the AI and they get an answer and then they go off and they never discover that they can say, build a Streamlit app and deploy it and host it and share it and all that sort of thing, right? So we're always looking for sort of signals, like when is it appropriate to, ⁓ you know, send somebody an email letting them know about a particular feature or... If somebody has been dormant for a couple of days, when should we send a, hey, we're still here kind of email, that sort of thing. If you were going to take a purely data focused approach to user activity and using the data to promote. ⁓ best use of your product. What are the like what are the outcomes? What are the things that you look at in order to to choose those points? Because you could just arbitrarily do it say ⁓ when they take action you know when they create a new code block then you just went black. You there?
Gaurav Jain: Yes, yes, I can hear you, but there's a problem. will come back. Sorry for this. Yeah, actually it's raining and. ⁓
Greg Michaelson: for the lights to add. All right, well, we'll just do it in the dark. ⁓ there they are. Nice.
Gaurav Jain: Yeah, the power generator just takes a few seconds to switch on. Sorry for that. Sorry. So you.
Greg Michaelson: Gotcha. No problem. That's fine. So I'm curious because part of me says, ⁓ I know the parts of the product and I know what are the right actions to take it sort of that sort of thing. But also there's the things that I don't know. So how do you how do you take a purely data driven approach to figuring out how to help your users find the best experience with the product? What is what are the optimization points? What are we looking for?
Gaurav Jain: Thank you. Right. Yeah. So like I said, right, so there are, so I mean, just to give you certain examples from, you know, the kind of customers we are working with, right? So for example, if... somebody have come and written something on your flag channel, right? Like for example, okay, I'm trying this product out and I'm facing this kind of a problem, right? Now what we have done is we have our own parsers and LLM. know, Intel extractors. And from there, we figured out, this person from this company is facing this sort of a problem. Now, if there are like four or five people from the same company have also been using this product in the last one month, right? Then we need...
Greg Michaelson: But how do you know it's four or five? What if it's three? What if it's seven? Like, how do you optimize it?
Gaurav Jain: I mean, yeah, yeah. So essentially ⁓ this comes from looking at the kind of product again, right? So for example, there are products where even two people can, you what we call as a activate the DQL, right? So, or there are products where, you know, even 10, 15 people or 15 developers are needed to bring that sort of an activation. For example, a company which is like, you know, you know, ⁓ a proper workflow orchestration sort of a company where a lot of developers work simultaneously on that product. Versus there is a company where maybe a people, know, an infra optimization sort of a company where two or three DevOps people are also good enough for you to start using that particular product. So depending upon those sort of things. But again, like I said, we are not configuring these things ourselves. As of now, we are letting the customer only come and define it. What we can do, I mean, if we have to do it, we can actually figure out by looking at your past activities, right, from the prospects which have become your customers, what was that activation number from developer standpoint, what kind of activities they have done, how much time they are spending on your documentation, and then start inferring from there to... You know the current prospect and how far are they from becoming your customer. So that could be one of the data approach which we can take, but we have not done that yet.
Greg Michaelson: So you wanna look at like say the money that they're spending, they're, you know, model that based on the different features that you have in your data, and then figure out the, optimize that spend.
Gaurav Jain: Right, so we know ⁓ the activities which are happening from their customers, right? How many developers are active from their customer? What is that threshold, right? ⁓ First for a company of, let's say this size to become their customer. So if we know that, we can actually start predicting that for the future prospects also.
Greg Michaelson: Got it. OK, well, that's that's certainly interesting. In the future, how are you going to do you have plans for how you're going to sort of optimize the approach and make it where the users don't have to do that sort of configuration? Because I imagine it changes over time as well. And you update.
Gaurav Jain: Yeah, I mean it does. Yeah, it does. mean, as the company grows, as the product grows, the kind of customers they are ⁓ targeting at, they're also, you things keep on changing. But I think the bigger problem which we are seeing, Greg, right now, specifically in this world is ⁓ with, you know, how agents are coming in, right? And... ⁓
Greg Michaelson: Okay.
Gaurav Jain: lot of time what is happening now is in fact some of the interesting patterns which we have seen in the last six months is the visits to documentations by developers have reduced or I would not say reduce significantly but for a very high growth startup it has started flattening up and for you know the companies which were
Greg Michaelson: What is started flattening out?
Gaurav Jain: the number of developers coming onto documentations, right? And ⁓ for companies which are pretty big, there what we have started seeing is it has actually started going down by 10, 15 % or even in some cases 20%, which is very, yes, yes. ⁓ yeah, so what has started happening and I'm pretty sure you would have seen now, ⁓
Greg Michaelson: ⁓ I see. Interesting. So people aren't reading documentation as much. what you're saying. Is that because?
Gaurav Jain: like some couple of months back there was this whole incident about tailwind, right, where the developers were not coming to their website or not coming to their documentation and they never got to know the entire, you know, the paid plan or the whole... pipeline of conversion was not happening because everybody was consuming the content through the AI or even inside the IDEs. So I think the biggest problem which needs to be solved today is to understand or first to acknowledge that ⁓ all of your assets as a DevTool company has to be machine readable and need to start working on agent experience because I mean believe it or not but developers are now delegating everything to a GCHS right. So, that is one thing and the second order problem for that now I mean from that now is given agents are discovering agents are evaluating. How do you get the intent which you were getting before? I like I said, developer intent was anyways very difficult earlier. Though we were able to solve it by getting data from a few places here and there and build it up. But with agents coming in, it is becoming much harder. So I think as... As we being in this industry, the one thing which we are trying to do is how do we bring agent intent to help you understand what they are trying to do, why they are trying to do. And that's where I think one of the biggest focus which we are working on right now is one, ⁓ building.
Greg Michaelson: Why is that hard? It seems like the easiest thing in the world. Like it makes it way easier if you can just see what they're asking for, doesn't it?
Gaurav Jain: Yeah, but capturing that is hard, right? I mean, how do you get to know that who is the person behind this agent? Sorry? Yeah, they do. ⁓ a lot of times, for example, if you think about documentations, what was happening right now, obviously, the identity was not available even in case of documentation on web.
Greg Michaelson: Why? I mean, companies control the agents. Don't the companies control the agents? Okay, yep.
Gaurav Jain: Right. But now that they are accessing the documentation through a cloud or a chat GPT or inside cursor, it's not possible for you to get, or I mean, it is possible, but you you are getting only a few intent or few questions which the developer is asking because that's where the LLM layer is coming in between. Right. So.
Greg Michaelson: ⁓ I see. So you've got users that are talking to JetGBT or Claude directly instead of through an agent built into your own platform. I see. Okay. I got it. That would be harder.
Gaurav Jain: So how do you start getting those sort of intent is something which we are trying to solve for. And second is also if you have your product MCP and people are calling it inside cloud, that's also something which we are trying to solve for. So these are the two things which we are solving for. The other thing which we are ⁓ now focusing on is trying to bring a lot of nuanced signals. For example, If somebody comes today and asks for, know, give me the list of companies who are migrating from, let's suppose, ⁓ Clickhouse, right? Or companies who have, you know, hired like, let's say these many DevOps in the last six months, right? also, a lot of these nuanced sort of parameters where it becomes very difficult for a normal, you know, LinkedIn based MCP to essentially answer to that. What we have done is we have ⁓ built a lot of knowledge on top of these databases and started giving that as an MCP. So I think these are some of the things which we are focusing on right now to, yeah, I think. help our customers to get those data points.
Greg Michaelson: Can you see when the large language models query the documentation for your customers?
Gaurav Jain: So, see the good thing is we don't need to get 100 % of it, right? What we need to do is if we can get, let's say 20, 30, 40 % of all the signals which are coming to us and if we can process it, figure out which companies these signals are coming from and present it to our customer, I think that's what we are targeting at right now. So yeah, we are able to do a decent amount right now. And yes, mean, that's something which we are working on to continuously improve on.
Greg Michaelson: But like you can see in the web traffic, can you tell, can you differentiate a regular visitor from a jet GPT?
Gaurav Jain: So in web traffic, yes, we can. Because ⁓ the moment you get the IP, you can obviously do it. ⁓ I think ⁓ in case of, let's say, example, a customer or if a developer is trying to use your product in their cursor or in their ID, ⁓ and if you have an MCP, about your documentation and if the developer has added that MCP in their ID is when we can start actually getting those signals. So yeah, I mean there are a lot of ifs right now, but yeah, the idea is that this ecosystem will keep on evolving and that's when these things will become a little better.
Greg Michaelson: Do you think so? I recently tried a project where I was trying to automate my grocery shopping and you know, I had a large language model that was going to figure out a meal plan for the week and then they were going to come up with recipes for each of the items of that meal plan and they know it knew what I had in my pantry in my fridge and then it would make a shopping list and. I sort of a wall there because the grocery stores, the brands, the websites, they block like bot traffic. And I think that we're gonna start seeing these online places.
Gaurav Jain: True.
Greg Michaelson: have to differentiate between bad bots and good bots. That's not really happening yet. Have you seen much of that?
Gaurav Jain: heading. So, see I am not very sure about Ecom, I though I have read a few ⁓ articles and I have not tried it by myself yet to do this sort of automation completely by cloud. But ⁓ in DevTool world it is happening for sure. In fact, ⁓ what we see is most of the DevTool companies today have exposed their documentation as an MCP. You know, rather than ⁓ expecting these ⁓ chat GPT or cloud to come and, you know, crawl it from your website, what they are doing is they are exposing it as an MCP and putting it inside and letting the developers put it inside the cursor or in their cloud connector, right? And this actually helps them reduce the hallucination as well. Also gives them the right information, right? rather than letting it decided by ⁓ one of these AIs to read it from multiple places and then present it, whatever they have understand. So the moment they present it as an MCP and add them inside your cloud or inside your cursor ⁓ is when the developer is also getting the right information. And I think that's how most of these different companies are solving for.
Greg Michaelson: So just provide an interface for the large language models to interact with rather than filtering out trying to try to allow some kind of bot traffic over other kinds of bot traffic.
Gaurav Jain: Okay.
Greg Michaelson: You think that's going to be the long-term ⁓ sort of method that this works with or is something going to come and replace MCP?
Gaurav Jain: I think so. I think this is making a lot of sense right now. mean think of it like you know how the whole Google search optimization and all those things evolved, right? How your results started coming in top when somebody was searching in Google. Same way, if you are giving your MCPs in Cursor or Cloud and everything, and if somebody comes and talk about, you I want to know about distributed, let's say SQL database, and suddenly, you know, it gives you the list of these four, five. tools and all of them are connected as part of your MCP. It is giving you the right parameters to evaluate on and yeah I think that's when the developer will also be able to make right decisions. So think this is going to evolve but I think this is the right path.
Greg Michaelson: All right, let's do long term view. I'd like to hear some predictions from you. What do we what do you see coming in your space in the next five years? I mean, it's risky even to predict five months at this point, but what do you see coming? Give me some predictions.
Gaurav Jain: So I agree. ⁓ fact, I don't know five years, but two, three things which are going to happen for sure. One is the... targeting will become more more nuanced. You will need better ICP classifications, right? And with AI coming in, that is becoming, I would say, possible now, right? So, yeah, so, right. So, you know, and in fact, before I jump into this, I'll probably talk about, you know, some of the things which make this problem very unique, right?
Greg Michaelson: Say more about me. I don't think I thought.
Gaurav Jain: So one of the things which I spoke about is developers are not leads, but they are the intent. I think we spoke about that. ⁓ The other very interesting problem is a lot of time, most of the companies come and define their ICP basis on industry, your basis on location, and that sort of a thing. But in DevTool, what happens is,
Greg Michaelson: you
Gaurav Jain: Technology, the fitment of technology becomes very, important. If I am, let's say, a company which only works with, or a library which works with Java, any company who is not using Java is not an ICP for me, right? So, you know, that sort of data points exactly. So, for example, we were working with this company which was... ⁓ JS library, right? And they have, I mean, it's big open source JS library. And what they have done is they have created a headless frontend for e-commerce engines, right? So you can take any e-commerce engine and then... ⁓ put a front end on top of that engine using that particular JS library. Now, for them to figure out whether this customer is qualified or not, they do not just need to look at the number of engineers, but they need to know how many front end engineers are there. And even in front end engineers, they need to know how many of them actually know this particular JavaScript or they can actually learn this particular JavaScript. So that sort of nuanced qualification is what is required. So I think that's what is going to happen right now. So that is probably the second thing which I wanted to talk about from an ICP standpoint. And third is, like I was talking about this whole agent. intent I think this whole industry will move from a beacon base or a JavaScript sort of a beacon which you install on your website it will move from there to more SDK based intent or like what we are talking about agent, MCP sort of a thing so I think these are the two three things which are going to happen in the next three to six months.
Greg Michaelson: So what's the, it's Rio.dev, R-E-O.dev.
Gaurav Jain: Yeah, I I did.
Greg Michaelson: What's next for you guys? Are you guys raising money? ⁓ What stage is the company at?
Gaurav Jain: Yeah, so we have raised C drum. We are in the process right now, mean more you will get to know pretty soon. But yeah, mean as of now we are a seed raised company. We have around 200 plus customers. big customers like NVIDIA or Temporal or CircleCI. So yeah, those sort of companies are working with us. We are also working with some new age AI companies like Langchain, CrewAI, Arise. So yeah, think our focus is going to be moving up market, closing more ⁓ bigger deals now. think we have got a few, think some 30, 40 customers who are like enterprise-date. We probably would want to increase that number. So yeah.
Greg Michaelson: Amazing. Well, hey, thank you so much for coming on. This has been really interesting learning about this space and learning about what you guys are doing. I'm excited to see where you go from here.
Gaurav Jain: Absolutely. Thanks.
Gaurav Jain: Absolutely great. Thanks for inviting me. It is amazing to be on data day with.
Greg Michaelson: With me. ⁓ Awesome. All right, well, hey, why don't we start out with an intro? Tell us who you are and what you're into and give me your background a little bit.
Gaurav Jain: So, ⁓ hey, I'm Gaurav. I am the CTO and co-founder at trio.dev. ⁓ In the industry for almost 18 years now, a couple of startups before this. At trio.dev, we are essentially solving a pretty interesting problem around how do you target ⁓ dev GTM. So, you can think of us as... ⁓ revenue intelligence platform trying to help companies that are selling to any kind of a technical audience, it a CTO or a CISO or ⁓ a VP engineering or architect or you know any of those kind of technical folks and I mean the problem which we are trying to solve is very simple reaching out to developers selling them has always been hard and don't come onto the calls. They'll block your ads. They'll never respond to your emails. So how do you basically make a sale? So that's what we are trying to do.
Greg Michaelson: Got it. Wow. That does sound like a hard problem. Certainly the amount of spam that's increased, you get on LinkedIn and over email has kind of gone through the roof.
Gaurav Jain: Absolutely, absolutely. fact, that is essentially the problem if you think it from a developer's standpoint how do we make sure that the selling is not annoying but it is more helpful for them. That is the developer side of this story.
Greg Michaelson: Okay, well, how did you get the idea? How did you get into that? You said you've been in the industry for 18 years? Talk more about your background.
Gaurav Jain: Yeah. Right. So, see, I have been, you know, all my life a technical person. I started my career with Amazon, one of the first few engineers to work on, work in Amazon India. In fact, one of the first few engineers here. Then very quickly jumped into a startup ecosystem. couple of startups before this one of the startup which I was doing in 2009. I tried it for two years didn't work out around 2016 or so is when I did another startup which is on basically we are trying to build a low code platform for digital lending so it was completely into fintech
Greg Michaelson: you Thank
Gaurav Jain: space did it for like six seven years sold it to multiple banks across india southeast asia middle east got an exit around 2022 took a break for a like almost seven eight nine months and then i was just trying to do something new this is when me my other two co-founders we sat together So what happened is one of my other co-founder who is currently the CEO at Trio, he in his previous startup, he was selling to developers. And one of the insight which he came up with was like, know, even though they had calls with CTOs or they were trying to build a pipeline at the end of the day, he was never clear of what my pipeline looks like because... you know, somebody who came and filled the form or had a good call with the sales people, nobody from the developer is actually trying out the product. Then after like, you know, maybe 15, 20 or 30 days when they are doing the call again, they realized that developers didn't like the product or they liked it a lot and they have made the decisions already. Right. So, so the sales team was not at all aware of what is really happening. And that's what the biggest problem they were facing when they were selling to developers. And when when Acyntia told me that this is the problem and then it became very easy for me to understand it because like I was saying, like 17, 18 years in the industry, I have purchased maybe 100, 150 plus tools. and hardly ever I would have gone on a demo call and made a decision because of the sales pitch. It was always based on what my developers are telling to me, how was their evaluation. So yeah, I think that's how we thought, okay, maybe this is an interesting problem to look at. We went ahead and spoke to like 100 plus. prospects, validated the idea, we understood, okay, this is a big problem for everyone. So that's how we started getting into it.
Greg Michaelson: So is it more of like a funnel management tool or is it more like how to break through spam or is it like ⁓ an outbound ⁓ like marketing tool or what does it do actually?
Gaurav Jain: So, think of it like ⁓ a three part, ⁓ you know, solution. So, one, we will help you figure out the data which is invisible, which is for example, you know, who are the people who are active on your GitHub repository, what kind of companies they belong to, who are the people who are. making some interactions with your Docker or your code command, package installers, or anything like that. Who are spending time on your docs? What kind of pages they are spending time on? So we try to bring in the data ⁓ of the developers plus the companies they are in. So that is one part. The second part is the Intel. part. So what we do is we build a developer qualified lead kind of a thing. I mean see you should I mean one of the thing which you always talk about is developers are not lead. developers are always intent. You should not reach out to developers the moment they come onto your website and fill a form. You should look at how much intensity, how many developers from the same company are coming, what is the time frame they are coming in. So what we have built on top of this data layer is an Intel layer where we are trying to build multiple scores, trying to bring in ⁓ the intent and the qualification part on this. And the third part is where we are giving you some sort of ⁓ automation or a workflow where you can actually define, okay, this happens, then this should happen, this is how we should reach out, what sort of messages we should reach out to them, when should we reach out to them. So, yeah, this is the solution.
Greg Michaelson: Got it. Where does the data come from?
Gaurav Jain: So we are collecting data from more than 30, 35 sources right now, starting from GitHub, Slack, Stack Overflow, Reddit, all those communities open data. We have beacons, which can be installed on your documentation. ⁓ get the first party data on your product to get the first party data. And then we also try to get data from package managers, be it your Docker, be it NPM, ⁓ PIP, all those sort of package managers. ⁓ In fact, what we have also started doing now, specifically ⁓ in this whole new world where developers are delegating their discoveries and evaluation to agents. We are also getting data from ⁓ MCPs ⁓ and there we are actually getting the intent in much, ⁓ I would say, a much friendly manner where people are writing that, okay, this is what we want to do. We are getting that. And so we are getting all the data from these places and then combining to give you.
Greg Michaelson: Right? Okay. Well, that's interesting. So, you mentioned that you shouldn't just reach out to developers the first time they fill out a form on your website. Talk to me about the logic that you guys have figured out around when to reach out to the folks that are signing up for a new product, So I sign up for a ton of products and I always get instant emails.
Gaurav Jain: Right, right. So see, I mean, there's obviously not ⁓ one formula which fits for all. But ⁓ I think there are multiple things which comes into the place, one is how many developers are coming and signing up on your product. I mean, for example, ⁓ somebody from Netflix has just come and signed up on your product. It could be just a hobby project for that person. I mean, if you're just reaching out to that person, it does not mean, fact, most of the time they'll get pissed off. Why are you reaching out to them? I'm not asking for any help. But let's say if, you know, six, seven or eight developers signed up in like two weeks or they are spending a lot of time on... docs, right, or specifically your docs which are talking about enterprise, production, deployment, or, know, security-related stuff. things like that plus if they have you know a lot of telemetry signals which are coming in. So, the moment you start triangulating all of these is when you see there is a good intent which is getting developed over there and that probably is the time when you should reach out you should not do it like the moment somebody just signed up. that is what we keep on trying.
Greg Michaelson: Got it. And do you guys build like a funnel for users? Like somebody signed up, somebody did action X, somebody did action X three times, that sort of thing. How do you develop that funnel? It's going to be different for every project.
Gaurav Jain: Right. So, funnel is very different from a typical MQL SQL sort of, ⁓ you know what happens in a B2B sort. So, what we say is a developer funnel looks like, you know, a four or five stage ⁓ process. First is... where a developer is discovering you, right? This is where they are just trying to know what you do, what kind of problems you are solving, whether it will be something good for me or not. The moment they start doing that evaluation, which is they are trying to figure out all the red flags, whether this works with my stack, does it work on the environment which I am working on. Do they have APIs or whatever the integration mechanism which I am comfortable with? So that's what we call as an evaluation step. So that is the second part. Once they have done the evaluation or they have figured out all the red flags and they know that, it's good to go from for us right now, they do what we call as a build a POC sort of a thing. So that's the third stage. No developer is ever going to. buy a product unless they have built something on top of your product. So a build a POC is a very, important stage there. That's where they are trying to ⁓ do some sort of package manager installation or they are spending a lot of time on your documentation. There are telemetry signals which are coming in. They are coming onto your Slack asking some questions which are giving you a hint, okay, they are evaluating or they are building something on top of my product. Once they have... done some POC, that's when it moves to what we call the fourth stage, which is a deployed stage. And this deployed stage is essentially where they would have probably taken a small plan and they are just using it for one particular use case. Two, three developers, they have deployed it, they are using it. And once they are confident is when they move to a scale stage. So yeah, this is a five stage funnel, we say discover, evaluate, build, deploy and scale.
Greg Michaelson: Got it. ⁓ so your functionality is mainly to sort of track users through those stages.
Gaurav Jain: Yeah, so what we do is essentially you get to know how many accounts are in which kind of a stage, right? When are they moving from one stage to another? What can you do to move a particular account from let us say evaluate to build a POC or from a build a POC to a deployed stage, right? And yeah, I think that is essentially what we do.
Greg Michaelson: How do you get to the what you can do kind of thing? Are you analyzing that data or is it a large language model query or how are you getting from you're in this stage and not that stage to here are the seven things you need to do to move someone from stage three to stage four.
Gaurav Jain: Right. two, three things which we do. One is, what we present is a timeline, activity timeline view, which essentially tells you, ⁓ the last three months, these many developers have done this, they've spent this much time on this kind of a documentation, then done this, this, this. So once you have all the data or all the activities which has happened from a particular account, ⁓ We also obviously get to know the location from where this is happening, the team they are working in. So then we can actually start recommending this, that, OK, now is the time when you should go and reach out to ⁓ these people. So we have our own models where you can just go and inquire, or you can research about an account and ⁓ shoot an email or. maybe reach out to them on LinkedIn with the problem they are trying to solve and what solution you can.
Greg Michaelson: But how do you know the optimum timing?
Gaurav Jain: So, see there are multiple ways and it depends upon you know the kind of tool also which is in the consideration right. For example, if it is database I mean you typically wait for a longer time I mean it does not happen so quickly, but when it is a simple API tool what we suggest is that for example, if somebody is visiting your pricing page after doing you know some that is a five six or seven types of activities. Maybe this is the time for you to reach out right or if somebody is spending a lot of time on let us say production deployment sort of a page. So, there are like you know multiple signals which essentially calls for an action right and once you get that signal you essentially take an action.
Greg Michaelson: But how do you know? mean, right. But every product is going to be different and all of the different activation points are are there.
Gaurav Jain: So we let the user define. Yeah, we let our server define. that's where the third layer comes in, right? So we have ⁓ all the granular intents or all the signals which are there, and then you can define the automations on top of it, right? So you can define things like, okay, five developers, pricing page with it, do this, right? Something of that sort.
Greg Michaelson: So the user has to divide this.
Gaurav Jain: Yeah, user can define. We do some recommendations, but at the end, like I said, a lot of ⁓ nuanced signals are there because there are a lot of these complex products which are there. So we let the user define those actions which they want.
Greg Michaelson: What if they don't know? Let's say this green field, I've got data of how users interact with my product on my website, and I'm interested in figuring out, you know, how should I market to those users to help them have the most success with my product possible? What kind of data elements are you looking at to identify those intervention points?
Gaurav Jain: Right, so ⁓ see I mean if you think about it there are like 100 plus signals right and what we have done is for each of these signals we have provided some sort of a weightage. For example, if you are a let's say a DevOps you know let's say Kubernetes cost optimization sort of a company right. So, what we have done is we have divided these products in these categories which could be either infra company or an API company or a AILM sort of a company like. And for each of these what we have done is we have built a layer of abstraction on top of it which is ⁓ if it is an optimization company look for these kind of signals, if it is an AI sort of a company look for these signals. So, what we do is we give them some assistance, but like I said at the end ⁓ every company has their own way to you know. build out these models. What we also provide is out of the other than the product, we also give ⁓ support from our forward deployment engineers who can actually come help them configure these automation workflows and build out good pipelines and all that. But yeah, at the end, mean, it's more of a consultative to figure out what works and what doesn't.
Greg Michaelson: Got it. That's kind of unsatisfying to me. I would love it if there was like ⁓ a data focused way to do that. Or do you have an agent built into ⁓ your product that can kind of dig through the data?
Gaurav Jain: ⁓ Yeah. So I think this is something which we have just launched a couple of months back. ⁓ We are trying to ⁓ build on top of it what we are calling it as, you so we have like different open data signals or real data signals as well. But we have not reached to a stage where we can go ahead and take it completely in our hand and start running the whole workflow by ourselves. And that's something which we want to do. like I said, there are a lot of nuances. mean, it becomes a little difficult to rather take care of all the nuances and give that human touch whenever the reach out is happening or outbound is
Greg Michaelson: So let's take Zerve for example. So my company is called Zerve. We're a development platform for researchers and analysts, people that are working with data. We have a coding agent baked into the product. And one of the things we see is that one of the things that we're working on right now is trying to help users find the different parts of the platform that they will take value from. There might be some users that
Gaurav Jain: you
Greg Michaelson: you know, come in and they ask a question to the AI and they get an answer and then they go off and they never discover that they can say, build a Streamlit app and deploy it and host it and share it and all that sort of thing, right? So we're always looking for sort of signals, like when is it appropriate to, ⁓ you know, send somebody an email letting them know about a particular feature or... If somebody has been dormant for a couple of days, when should we send a, hey, we're still here kind of email, that sort of thing. If you were going to take a purely data focused approach to user activity and using the data to promote. ⁓ best use of your product. What are the like what are the outcomes? What are the things that you look at in order to to choose those points? Because you could just arbitrarily do it say ⁓ when they take action you know when they create a new code block then you just went black. You there?
Gaurav Jain: Yes, yes, I can hear you, but there's a problem. will come back. Sorry for this. Yeah, actually it's raining and. ⁓
Greg Michaelson: for the lights to add. All right, well, we'll just do it in the dark. ⁓ there they are. Nice.
Gaurav Jain: Yeah, the power generator just takes a few seconds to switch on. Sorry for that. Sorry. So you.
Greg Michaelson: Gotcha. No problem. That's fine. So I'm curious because part of me says, ⁓ I know the parts of the product and I know what are the right actions to take it sort of that sort of thing. But also there's the things that I don't know. So how do you how do you take a purely data driven approach to figuring out how to help your users find the best experience with the product? What is what are the optimization points? What are we looking for?
Gaurav Jain: Thank you. Right. Yeah. So like I said, right, so there are, so I mean, just to give you certain examples from, you know, the kind of customers we are working with, right? So for example, if... somebody have come and written something on your flag channel, right? Like for example, okay, I'm trying this product out and I'm facing this kind of a problem, right? Now what we have done is we have our own parsers and LLM. know, Intel extractors. And from there, we figured out, this person from this company is facing this sort of a problem. Now, if there are like four or five people from the same company have also been using this product in the last one month, right? Then we need...
Greg Michaelson: But how do you know it's four or five? What if it's three? What if it's seven? Like, how do you optimize it?
Gaurav Jain: I mean, yeah, yeah. So essentially ⁓ this comes from looking at the kind of product again, right? So for example, there are products where even two people can, you what we call as a activate the DQL, right? So, or there are products where, you know, even 10, 15 people or 15 developers are needed to bring that sort of an activation. For example, a company which is like, you know, you know, ⁓ a proper workflow orchestration sort of a company where a lot of developers work simultaneously on that product. Versus there is a company where maybe a people, know, an infra optimization sort of a company where two or three DevOps people are also good enough for you to start using that particular product. So depending upon those sort of things. But again, like I said, we are not configuring these things ourselves. As of now, we are letting the customer only come and define it. What we can do, I mean, if we have to do it, we can actually figure out by looking at your past activities, right, from the prospects which have become your customers, what was that activation number from developer standpoint, what kind of activities they have done, how much time they are spending on your documentation, and then start inferring from there to... You know the current prospect and how far are they from becoming your customer. So that could be one of the data approach which we can take, but we have not done that yet.
Greg Michaelson: So you wanna look at like say the money that they're spending, they're, you know, model that based on the different features that you have in your data, and then figure out the, optimize that spend.
Gaurav Jain: Right, so we know ⁓ the activities which are happening from their customers, right? How many developers are active from their customer? What is that threshold, right? ⁓ First for a company of, let's say this size to become their customer. So if we know that, we can actually start predicting that for the future prospects also.
Greg Michaelson: Got it. OK, well, that's that's certainly interesting. In the future, how are you going to do you have plans for how you're going to sort of optimize the approach and make it where the users don't have to do that sort of configuration? Because I imagine it changes over time as well. And you update.
Gaurav Jain: Yeah, I mean it does. Yeah, it does. mean, as the company grows, as the product grows, the kind of customers they are ⁓ targeting at, they're also, you things keep on changing. But I think the bigger problem which we are seeing, Greg, right now, specifically in this world is ⁓ with, you know, how agents are coming in, right? And... ⁓
Greg Michaelson: Okay.
Gaurav Jain: lot of time what is happening now is in fact some of the interesting patterns which we have seen in the last six months is the visits to documentations by developers have reduced or I would not say reduce significantly but for a very high growth startup it has started flattening up and for you know the companies which were
Greg Michaelson: What is started flattening out?
Gaurav Jain: the number of developers coming onto documentations, right? And ⁓ for companies which are pretty big, there what we have started seeing is it has actually started going down by 10, 15 % or even in some cases 20%, which is very, yes, yes. ⁓ yeah, so what has started happening and I'm pretty sure you would have seen now, ⁓
Greg Michaelson: ⁓ I see. Interesting. So people aren't reading documentation as much. what you're saying. Is that because?
Gaurav Jain: like some couple of months back there was this whole incident about tailwind, right, where the developers were not coming to their website or not coming to their documentation and they never got to know the entire, you know, the paid plan or the whole... pipeline of conversion was not happening because everybody was consuming the content through the AI or even inside the IDEs. So I think the biggest problem which needs to be solved today is to understand or first to acknowledge that ⁓ all of your assets as a DevTool company has to be machine readable and need to start working on agent experience because I mean believe it or not but developers are now delegating everything to a GCHS right. So, that is one thing and the second order problem for that now I mean from that now is given agents are discovering agents are evaluating. How do you get the intent which you were getting before? I like I said, developer intent was anyways very difficult earlier. Though we were able to solve it by getting data from a few places here and there and build it up. But with agents coming in, it is becoming much harder. So I think as... As we being in this industry, the one thing which we are trying to do is how do we bring agent intent to help you understand what they are trying to do, why they are trying to do. And that's where I think one of the biggest focus which we are working on right now is one, ⁓ building.
Greg Michaelson: Why is that hard? It seems like the easiest thing in the world. Like it makes it way easier if you can just see what they're asking for, doesn't it?
Gaurav Jain: Yeah, but capturing that is hard, right? I mean, how do you get to know that who is the person behind this agent? Sorry? Yeah, they do. ⁓ a lot of times, for example, if you think about documentations, what was happening right now, obviously, the identity was not available even in case of documentation on web.
Greg Michaelson: Why? I mean, companies control the agents. Don't the companies control the agents? Okay, yep.
Gaurav Jain: Right. But now that they are accessing the documentation through a cloud or a chat GPT or inside cursor, it's not possible for you to get, or I mean, it is possible, but you you are getting only a few intent or few questions which the developer is asking because that's where the LLM layer is coming in between. Right. So.
Greg Michaelson: ⁓ I see. So you've got users that are talking to JetGBT or Claude directly instead of through an agent built into your own platform. I see. Okay. I got it. That would be harder.
Gaurav Jain: So how do you start getting those sort of intent is something which we are trying to solve for. And second is also if you have your product MCP and people are calling it inside cloud, that's also something which we are trying to solve for. So these are the two things which we are solving for. The other thing which we are ⁓ now focusing on is trying to bring a lot of nuanced signals. For example, If somebody comes today and asks for, know, give me the list of companies who are migrating from, let's suppose, ⁓ Clickhouse, right? Or companies who have, you know, hired like, let's say these many DevOps in the last six months, right? also, a lot of these nuanced sort of parameters where it becomes very difficult for a normal, you know, LinkedIn based MCP to essentially answer to that. What we have done is we have ⁓ built a lot of knowledge on top of these databases and started giving that as an MCP. So I think these are some of the things which we are focusing on right now to, yeah, I think. help our customers to get those data points.
Greg Michaelson: Can you see when the large language models query the documentation for your customers?
Gaurav Jain: So, see the good thing is we don't need to get 100 % of it, right? What we need to do is if we can get, let's say 20, 30, 40 % of all the signals which are coming to us and if we can process it, figure out which companies these signals are coming from and present it to our customer, I think that's what we are targeting at right now. So yeah, we are able to do a decent amount right now. And yes, mean, that's something which we are working on to continuously improve on.
Greg Michaelson: But like you can see in the web traffic, can you tell, can you differentiate a regular visitor from a jet GPT?
Gaurav Jain: So in web traffic, yes, we can. Because ⁓ the moment you get the IP, you can obviously do it. ⁓ I think ⁓ in case of, let's say, example, a customer or if a developer is trying to use your product in their cursor or in their ID, ⁓ and if you have an MCP, about your documentation and if the developer has added that MCP in their ID is when we can start actually getting those signals. So yeah, I mean there are a lot of ifs right now, but yeah, the idea is that this ecosystem will keep on evolving and that's when these things will become a little better.
Greg Michaelson: Do you think so? I recently tried a project where I was trying to automate my grocery shopping and you know, I had a large language model that was going to figure out a meal plan for the week and then they were going to come up with recipes for each of the items of that meal plan and they know it knew what I had in my pantry in my fridge and then it would make a shopping list and. I sort of a wall there because the grocery stores, the brands, the websites, they block like bot traffic. And I think that we're gonna start seeing these online places.
Gaurav Jain: True.
Greg Michaelson: have to differentiate between bad bots and good bots. That's not really happening yet. Have you seen much of that?
Gaurav Jain: heading. So, see I am not very sure about Ecom, I though I have read a few ⁓ articles and I have not tried it by myself yet to do this sort of automation completely by cloud. But ⁓ in DevTool world it is happening for sure. In fact, ⁓ what we see is most of the DevTool companies today have exposed their documentation as an MCP. You know, rather than ⁓ expecting these ⁓ chat GPT or cloud to come and, you know, crawl it from your website, what they are doing is they are exposing it as an MCP and putting it inside and letting the developers put it inside the cursor or in their cloud connector, right? And this actually helps them reduce the hallucination as well. Also gives them the right information, right? rather than letting it decided by ⁓ one of these AIs to read it from multiple places and then present it, whatever they have understand. So the moment they present it as an MCP and add them inside your cloud or inside your cursor ⁓ is when the developer is also getting the right information. And I think that's how most of these different companies are solving for.
Greg Michaelson: So just provide an interface for the large language models to interact with rather than filtering out trying to try to allow some kind of bot traffic over other kinds of bot traffic.
Gaurav Jain: Okay.
Greg Michaelson: You think that's going to be the long-term ⁓ sort of method that this works with or is something going to come and replace MCP?
Gaurav Jain: I think so. I think this is making a lot of sense right now. mean think of it like you know how the whole Google search optimization and all those things evolved, right? How your results started coming in top when somebody was searching in Google. Same way, if you are giving your MCPs in Cursor or Cloud and everything, and if somebody comes and talk about, you I want to know about distributed, let's say SQL database, and suddenly, you know, it gives you the list of these four, five. tools and all of them are connected as part of your MCP. It is giving you the right parameters to evaluate on and yeah I think that's when the developer will also be able to make right decisions. So think this is going to evolve but I think this is the right path.
Greg Michaelson: All right, let's do long term view. I'd like to hear some predictions from you. What do we what do you see coming in your space in the next five years? I mean, it's risky even to predict five months at this point, but what do you see coming? Give me some predictions.
Gaurav Jain: So I agree. ⁓ fact, I don't know five years, but two, three things which are going to happen for sure. One is the... targeting will become more more nuanced. You will need better ICP classifications, right? And with AI coming in, that is becoming, I would say, possible now, right? So, yeah, so, right. So, you know, and in fact, before I jump into this, I'll probably talk about, you know, some of the things which make this problem very unique, right?
Greg Michaelson: Say more about me. I don't think I thought.
Gaurav Jain: So one of the things which I spoke about is developers are not leads, but they are the intent. I think we spoke about that. ⁓ The other very interesting problem is a lot of time, most of the companies come and define their ICP basis on industry, your basis on location, and that sort of a thing. But in DevTool, what happens is,
Greg Michaelson: you
Gaurav Jain: Technology, the fitment of technology becomes very, important. If I am, let's say, a company which only works with, or a library which works with Java, any company who is not using Java is not an ICP for me, right? So, you know, that sort of data points exactly. So, for example, we were working with this company which was... ⁓ JS library, right? And they have, I mean, it's big open source JS library. And what they have done is they have created a headless frontend for e-commerce engines, right? So you can take any e-commerce engine and then... ⁓ put a front end on top of that engine using that particular JS library. Now, for them to figure out whether this customer is qualified or not, they do not just need to look at the number of engineers, but they need to know how many front end engineers are there. And even in front end engineers, they need to know how many of them actually know this particular JavaScript or they can actually learn this particular JavaScript. So that sort of nuanced qualification is what is required. So I think that's what is going to happen right now. So that is probably the second thing which I wanted to talk about from an ICP standpoint. And third is, like I was talking about this whole agent. intent I think this whole industry will move from a beacon base or a JavaScript sort of a beacon which you install on your website it will move from there to more SDK based intent or like what we are talking about agent, MCP sort of a thing so I think these are the two three things which are going to happen in the next three to six months.
Greg Michaelson: So what's the, it's Rio.dev, R-E-O.dev.
Gaurav Jain: Yeah, I I did.
Greg Michaelson: What's next for you guys? Are you guys raising money? ⁓ What stage is the company at?
Gaurav Jain: Yeah, so we have raised C drum. We are in the process right now, mean more you will get to know pretty soon. But yeah, mean as of now we are a seed raised company. We have around 200 plus customers. big customers like NVIDIA or Temporal or CircleCI. So yeah, those sort of companies are working with us. We are also working with some new age AI companies like Langchain, CrewAI, Arise. So yeah, think our focus is going to be moving up market, closing more ⁓ bigger deals now. think we have got a few, think some 30, 40 customers who are like enterprise-date. We probably would want to increase that number. So yeah.
Greg Michaelson: Amazing. Well, hey, thank you so much for coming on. This has been really interesting learning about this space and learning about what you guys are doing. I'm excited to see where you go from here.
Gaurav Jain: Absolutely. Thanks.
