Rahul Pattamatta - CEO at DataBrain

Welcome to another CEO Wisdom Podcast. We have Rahul Patamata with us today. Rahul is a co-founder and CEO at Databrain, uh, an agent analytics platform. Started that company, uh, 4 years ago, got funded, and is scaling from our pre-conversations. Excited to have him on. Rahul, tell us a bit more about yourself and what you're building.

Yeah, thanks, Charles, firstly for having me. Uh, super excited. To chat with you. I had a chance to look at some of your guests and like, very cool, the kind of folks you're getting on. So a bit about me. So I, I'm an engineer at heart, uh, been a backend engineer, worked a lot on like distributed systems and machine learning products before machine learning was cool. I went to grad school, uh, and, uh, in machine learning. So always a big fan of data and infrastructure. Uh, I spent about a decade in in Silicon Valley working across a couple of startups, learned a lot on like, you know, building and shipping product. And also like one thing that Valley does really well is like kind of the feedback loop. Once you ship a product, kind of measuring and iterating and seeing how things work and then kind of improving product. So after a decade of working, I caught the startup bug. I always wanted to start something. So at some point I realized I wanted to, take the deep dive and start a company around the customer success space. Made a lot of first-time founder mistakes, especially around building like very cool product and then thinking about trying to fit that into customers' problems. And that kind of led me to Databrain. And Databrain was founded like 4 years ago. And essentially we're like an agentic analytics platform. We're rewriting the data stack for analytics and BI with the whole Agentic wave. And yeah, we're super excited to be building in the space. I'm very passionate about, like I said, like data and infra, so.

So how did you scale to where you are today in these 4 years? Do you have like, yeah, I don't know, a top 3 of how to access customers and to retain them?

Yeah, great question. So I think it's obviously never a linear path. There's a lot of zig and zagging. A lot of startup stories sound cool on the outside, but like on the inside it's just like a lot of hard work and like eating glass. We initially started off as a point solution in the data space, Semanticlayer. So one of the mistakes we wanted to avoid was to not build, I mean, like to not I mean, build early, we wanted to like kind of focus on talking to a lot of customers and that's essentially what we did initially. And then that led us to pivoting our product slightly, not into a completely different space, but like in the data space. And then like we essentially did a lot of cold outreach to a lot of people on LinkedIn. It really helps if you have a background in a particular area in like a specific,, you know, area of expertise. Like, so I, since I worked in the data space and as an engineer, people were like super receptive to, you know, giving feedback and talking about ideas. Um, and you don't necessarily want to do hard selling at that point. You just want to like chat with people about like new things that are out there in the market and, and, uh, and kind of build up like the trust. Um, and, uh, and the goal was to get someone who wanted to take a bet on us and, and like help us build our roadmap and become like our design partner. So that's what we did initially. We spoke to a lot of folks on LinkedIn and also used my network in the data space and reached out to people who, who are looking for like a solution like this and had problems in this space.

And what is that problem, right? And how is your solution the right one to fix it in that sea of other options?

Yeah, so I think most SaaS companies, I know SaaS now is kind of like, you know, under threat because of AI, but like historically, if you look, most SaaS companies when they scale, they reach a point where security of their data becomes really important. So like even before we get to that, I think what happens when they reach a certain scale is that like a lot of their users generate data on their platform by virtue of workflows, right? So if you can think of like a contract management platform or like a procurement SaaS company, their procurement-related workflows that their users do, like a procurement analyst does, or if it's like a contract management company, then like a legal analyst does, all this data gets generated and sits in like a database somewhere. And typically the users come to the product manager and they're like, hey, I'm generating all this data, I'm doing all these things, I would love to look at this from like an admin perspective or like an ops perspective and see how I'm doing and then like what the analytics are and like what the metrics are. And at that stage, um, product managers have two things. They can either build this in-house or they can go look for a solution like Dataprint. And the reason why we even exist today is because it's hard to build security and governance and like a lot of these engineering, hard engineering, uh, uh, problems, um, but for engineers, right? Like, so you have to think about data leaking from one tenant to another tenant. You have to think about like within a particular tenant, like which users get to see what kind of data, what the visualizations are. A lot of our competitors who were small were focused a lot on like the easy challenges, like really nice visualizations and UI. Whereas we went after like the hard problems. We started like thinking about the real problems that engineering teams face, primarily because we were engineers ourselves., and we kind of like started talking to those companies.

Selling to engineers is notoriously hard. Um, do you still cold email them? Do you still send them LinkedIn message? What does your messaging look like? How do you address these very introverted people that do not react easily to sales rhetoric?

Yeah, yeah, no, great question again. I think we actually are used by engineers, but the decision maker is like product folks, product managers, because it's in their roadmap. And surprisingly, product people are very social. They love talking to us, they love talking to, taking demos and seeing things. So the way our GTM kind of works is like, Engineers, we have to build product for engineers to really love us, but they don't take calls, they don't take meetings, they sign up for the product. So we have like a PLG workflow where they self— they sign up themselves, they self-serve. We got to impress them with like really good user experience and developer experience, solid developer docs. And then the product managers care about slightly fewer different things, right? Like, so they care about like shipping fast and making sure their roadmap is is sorted out and things like that. So we had to, we handle both these things today. And like you said, if you only focus purely on engineering, it's really, really hard, right? Like it's sometimes they like, they care about open source solutions. They just like, you know, some of the engineers think about like wipe coding solutions. They don't necessarily think through like the hard problems and challenges and things like that. So it is a challenge.

And why not founders or CEOs? Aren't these faster buyers, lower sales cycles and product managers managers that have less buying responsibility/accountability?

Yeah, so the thing is, to be honest, you're right about all those things about founders, but eventually they turn out to be like bad long-term customers because a lot of companies die.

But founders with product managers under them is my point.

Yeah, that's a good point. It's something that we haven't actually evaluated, but like We have sort of shifted our focus a lot to mid-market and enterprise. So like if you reach out to a founder of like a 500-person company, there's no way they're gonna like even, it's gonna go to their spam because they get hit up so much and they're like sort of like PMs. So we directly reach out to like, you know, the product management leads and stuff.

Right. And the pain point is that some of these folks think they can build their own dashboards, but in reality it's incredibly hard. It's technical, they need to plug a bunch of source of data and it's like hell and they end up like not building anything for 6 months and then people are just pissed that they have their eyes on different goals and different priorities.

Yeah, absolutely, you hit the nail on the head. And then even if you go to competitive SaaS industries where they're like 3 to 4 players, their users get frustrated with poor reporting, go to G2 or Capterra and leave a bad review about their reporting. That's when we reach out to a lot of these guys and be like, "Hey, look, you have to step up your reporting. It's not the greatest in the world right now." So.

Agentic reporting, right? Like me as a CEO, yes, I like to have my brain on some data and digest it, very important pieces. And to be frank, I don't even know, right? Like I've been in business for like 15 years and yeah, if you ask me, like I want to know how many leads I have in the pipe, the value of these leads, right? The close rates, my churn rates, of course my MRR, but it's hard to know like what you want to, where you want to have your brain look at, right? So that's one thing that I feel I would need help with. I do have some technical folks in my team that are building my own dashboards and I do have some SaaS as well that have like reporting capability, but do you also provide help with like knowing what to show to the end user?

Yeah. So we do have a— we do plug into your database and typically show a couple of things. One is proactively show you what you need to track. And then there's the reactive piece where you know exactly what to track and you kind of talk to the database and sort of come up with your own insights and stuff. So there's a mix of both that happen. But like, you know, Easier than it sounds, you know, a lot of this stuff is still in there, but it's not like tracked at all. And it's one of those things where it's like, I'm trying to find a good analogy, right? Like, it's, I mean, the very broad analogy is like electricity. You take it for granted, it's always there, and you're like kind of, you know, but when it goes down, you're like, oh damn, I'm working, it needs to. So a lot of like KPIs and metrics are very similar. You sort of have like all these numbers and everything. That you track. There are a few types of companies that track these religiously on a day-to-day basis, but most of them just like have it in the background. But then when like things break, they better— they really need to know like what's happening and why things fail and things like that, right? So we're getting to that world. I think the sort of shape the industry is taking is it's following very similar patterns to how like software engineering is evolving. Initially when Cursor, for example, came out, software engineering was all about hitting tab and autocompleting, right? And then we went from that to sort of creating one agent and that kind of does stuff and then you're going editing the agent and giving feedback. Now engineers don't even look at the code. There's agents and there's subagents and there's work trees. Data work is following a similar pattern too. Because everything is getting decomposed, you might have this question in your head. Charles wants to know revenue and then you spin off an agent. You don't bother an engineer or you go to a data person and say, hey, you only have one person on your team who does that. And they write this long-running agent that does a bunch of things. They get a feedback from the agent saying, hey, I need to know what revenue actually means in your company. And there's like the human in the loop. So we're following the exact same pattern, and, uh, in effect what we're building is the infrastructure for to enable that for like customers.

Got it. So what do I get for $2K per month versus $1K per month?

Yeah, so, uh, the entry point, by the way, which we are, which we are changing, uh, we're dropping it closer to Claude because a lot of the feedback was, uh, a lot of the feedback was that like, hey, you know, very small companies can like kind of build a very simple static dashboard solution. The sort of $2K per month is when you're kind of scaling, you're thinking of buying a real analytical solution and you don't want the overhead of managing it yourself, the infrastructure, and you want to go to production with maybe a handful of customers, get real feedback, but you don't necessarily want the— you get a cloud version, but you don't necessarily get the entire guardrails of like self-deployment, right? Like a lot of our enterprise customers don't want any of the data to leave their VPC. It runs inside their private cloud. So you don't necessarily get that in, so it's like for sort of like companies that are graduating from SMB to mid-market.

Interesting. And where I was trying to get with that other question, so let's say I pay you, I get some consulting, I get some software, of course, What's the agentic part, right? Because as a CEO, if I can have agents that fetch the data plus take action on it, like that's the, that's paradise, right?

So yeah, yeah, that's exactly, uh, that's exactly sort of like what we're building, right? Like, so the idea is that, um, you know, you, you think of, uh, the similarity between like your code and data. You want some insights out of it. You have all these questions in your head and you want to get some— something done, right? Like, for example, earlier I was trying to build out, um, I identify all the unstable parts of our product from our logs. And this is a data problem. You have a bunch of logs and sitting somewhere in a data lakehouse. Um, and I wrote this query and then it like spawned up sub-agents. It tried to understand like, uh, what errors are. So it took like 10 minutes to kind of go and do a different kind of work. Typically that would take like a day or two for like a data analyst to do. And we're compressing that down to like, you know, minutes of work. And we're graduating from like simple fact finding, like, you know, which were the previous generation of BI and reporting, which is tell me what, like, you know, what happened to my revenue, what is my revenue last month or like last 6 months to now, why is my revenue dropping and what can I do about it? Did I launch something that affected my revenue, right? Like so very, very deep questions that ingest a lot of the context in the organization that, so you're going from like the what to the why kind of questions.

What's the confidence rate on that? Do you take like real-time full market data? Because sometimes my revenue might, be down not because of my own lack of efforts or a broken system, but the state of the market, right? So do you also analyze external factors and real-time factors like that?

Yeah, we do have integrations. We don't automatically ingest right now. We do have integrations where we're working on a couple of integrations with Parallel, um, where you can ingest external data. But a lot of the requests for our existing customers is like ops and internal data related. Which is things that they already kind of have stored somewhere. But you're right, it's not like, it doesn't cover the entire 100% of the surface area. There could be some externalities, yeah.

Give me examples of softwares that you use in business and personally in which you love the analytics, right? I'll start on my end, Eight Sleep, right? Like my sleep score is great, it's like super clear and I can scroll down and I think their dashboard's amazing. Uh, Whoop maybe, although Whoop is like less accurate. Um, and then obviously like Stripe in business, but I, I think even Stripe is confusing, right? It's like, I don't— there's too many things. I feel they made the product for developers. Um, and I sometimes I just don't know where to click, you know, like very simple stuff like how much I made this month and stuff that is relevant, right? Like I feel some AI should scrape all of it and give me insights. And, you know, like just a quick insight light bulb section. Hey Charles, did you know that blah blah blah. Um, what about you? What products do you use that have like amazing dashboards and data are super useful to you?

Yeah, so I'll name them, but I also will talk about the counterpoint because I had a conversation with a data leader earlier, uh, earlier today, and basically they want to use us for like They're a very large bank and they want to use us for like, they have dashboards and they want to use us for like summarizing.

Banks are usually the worst when it comes to giving useful data. I mean, except like how much is there in their bank account, like they're the worst.

Yeah, exactly. Like it's, it's, uh, it's really poor. I think you covered a lot of the, my favorite ones by Whoop. I think I really like, although, um, you know, you kind of obsess about like, there's this concept called vanity metrics where you obsess about the wrong thing. Sometimes Whoop can potentially give you that. Like, I think that's where I'm super excited about this insight and summarization because you can find the needle in the haystack because having that AI, right, that you can ask questions to, I think that's pretty smart.

It's like everyone's on to ChatGPT and you just ask questions to that AI.

Have you thought about that UX factor where you just talk to an AI like a chatbot?

Well, you have a dashboard, but right next to it you have a bubble and you can just ask any question to an AI about your data. It seems so simple.

Yeah, yeah, yeah, we've thought, we've spoken about that. We haven't launched it, but it's something that we, we, uh, we've definitely spoken about internally. Uh, we have a completely clean window with like a chatbot where you can talk to the database, but then like the next evolution of that is like talking to the dashboard itself, which like you said, um, which I think is like super cool.

What other products? Give me other examples.

Yeah, like honestly, we deal like day to day with a lot of like, uh, SaaS companies. I can name a few, like boring but really useful ones in the enterprise space, right? So you got a procurement tech company that sells commodity software, very, very like doing extremely well. And basically they have this history of 20 years of data that they want to kind of productize and make it look aesthetically very nice to heads of procurement.

What about businesses that their full livelihood depends on dashboards like YouTube Ads for YouTube for creators, right? My YouTube channel all the time. It's pretty good. Um, have you analyzed that one, or AdWords, or Meta Ads, for example? Do these dashboards look good?

Um, actually, yeah, as you're talking about it, Ahrefs is something that I obsess about, actually. Uh, Semrush, Ahrefs, uh, Google Analytics is pretty good too, actually. Pretty solid. It hasn't changed, but it's like kind of gets the job done. Actually, that's a really good one that I like. Google Analytics.

What's the mental exercise that you do to know if a dashboard is good for the end user?

Yeah.

Their feedback, do you track their movements? How do you know?

Yeah, we don't, we should be.

Do you track their eyes?

We don't, that's too creepy. We actually get on calls with them because like a lot of it is very subjective. It's like art to be very honest when it comes down. There is like sort of the base defaults which have to be really good.. But then like beyond that, it's like so subjective. You can look at it and be like, I don't understand anything out of this. Someone else may be like, hey, I'm very quant, I've been like used to charts and dashboards since high school and I get this. There is a stat somewhere that I saw that like, you know, I would say like over 80, 85%, crazy number, like cannot read like a simple bar chart. Like it's too intimidating for people. And we see that the whole industry obsesses about dashboards and charts, but like people just, like find it, still find it very intimidating. So like having AI insights and summaries next to it, uh, actually helps a lot.

There you go. What do you have on your roadmap for this year? What do you want to accomplish?

Yeah, so, uh, we are launching the Agentic Analytics, like, like the, the actual infrastructure, a long-running agent. So we're going to be doing like sandboxes, um, very similar to what like Cursor does, like for like cloud agents. Uh, we're going to be launching like sandboxes where you can run long-running agents for data analysis and data science work. So it does both SQL and Python. We are super excited about that. We are changing our reskin of our product, which is coming out very, very soon. A ton of stuff lined up in the MCP. There's a CLI that we're going to launch for developers, CLI tool, so they can do a lot of the semantic layer stuff, setting up the prerequisites for AI through the CLI. So yeah, a bunch of these lined up and plenty more.

Now as a CEO, what does your day-to-day look like? Is it, I don't know, 25% managing coders, 25% GTM, 25% retaining customers? What does it look like?

Yeah, it's roughly around that, but it varies on a day-to-day because there might be a hair-on-fire problem. Customer might be like, "Hey, I got this big demo and you gotta help us out.". And sometimes it gets escalated, I gotta get involved. And I try and write code to stay sharp. I mean, although I'm not writing code like I used to because with AI, you're kind of prompting. And so I would say managing developers is reducing slowly because they're getting on very clear roadmaps and on autopilot mode. But a lot of my time is spent on GTM. Making sure like we're writing the right content, like doing outbound stuff, getting on demos with customers, getting feedback from customers, retaining them and stuff. So I would say like 40%, 50% on GTM stuff, 20%, 30% on like customer retention, and then the remaining on like engineering and like ad hoc stuff, like sometimes finance and like, you know.

And coding for you nowadays, what does it look like? Is it cloud code mostly and how do you interact with it?

Yeah. Cloud Code through the CLI. I'm also a Cursor ambassador, so I use a lot of Cursor. I love Cursor.

There's almost no one using Cursor nowadays.

Yeah, well, surprisingly, the enterprise has a lot of— we live in a bubble of Twitter, right? We see a lot of startups and sub-5, 10-employee companies. Even on Cursor. Yeah. If you go to enterprises, it's all like Cursor, to be very honest with you. Like, Cursor cracked the GTM, uh, really well. Of course, there's like Cloud— they use Cloud Code too.

Yeah, they made— they were acquired for $60 billion, right? So they were probably making what, like $5 billion a year, like when they got acquired?

Yes, they were doing really well. Yeah, they were doing really well. And they cracked—

who other than enterprises to pay that, right? It's like, yeah, there's only enterprise that can pay that level.

And it's a phenomenal, uh, example of like If you go to their website from archive, web archive from like 3 years ago when they launched, just imagine. And like early days, very, very early days, very clear on the GTM for like large enterprises. And so they built product based on that. So they indexed large files. They were the best in indexing. And, you know, Cloud Codecs both didn't focus on indexing. They sent the entire codebase as tokens and stuff. So like these small things that they did made a huge difference.

Yeah, and tracking the enterprise code GTMY is incredibly hard. Is that one thing you're trying to do on your end?

Yes, yes.

And so what's your strategy for that? Because you, I mean, chewing glass, I think, is the number one, because like you need to wait for a very long time and bow down to their silly, very silly requests, right? Um, so how do you deal with that?

Yeah, it's really hard, but it's worth it, um, because obviously like you gotta charge equal amounts for it and cash flow is good. But like there will be times when like you don't see any hope or any light. It's really tough, long sales cycles. But the way to actually do it is, to be very honest, it's only like 50 or 60% product. A lot of it is like relationships and like kind of, you know, giving the trust to enterprises.

Yeah.

Navigating the politics and like giving trust.

Yeah, and sometimes you pay out of pocket, right? It's like entering a Costco with a physical product. You need to have very deep pockets and patience., and, but yeah, once the gates flood open, not only is it the gigantic sum of money, but there the churn is like almost nonexistent because these guys love comfort.

Yeah, yeah, yeah. And they don't want to go through the procurement process again, right? So, right, right, exactly. 3, 4-year contracts. And, and, uh, and the good thing is like your roadmap is sorted, cuz like you build, you have to figure out, that's where like judge, good judgment comes in. Like you have to figure out like, are they asking silly things or are they asking things that will affect other customers of mine and improve their experience, right?

Yeah, worst case can branch a specific version for them, but that's headaches as well.

Yeah.

Where can people find out more about you, Rahul?

Yeah, so I'm on X, rpatnam. I'm on LinkedIn, Rahul Padamatta on LinkedIn. These are two places mostly. I talk on LinkedIn a lot about like data stuff. I'm gonna increase my visibility there and X on like random startup stuff.

So now this is Rahul and Charles