← BACK TO PROFILES

Vikram Chennai

Founder and CEO·Ardent·San Francisco, California·

Sell first, then build what people already want

In this interview

In this interview, Vikram Chennai, founder and CEO of Ardent, explains why coding agents stall at the database and how copy-on-write cloning gives every agent a full production Postgres copy in under six seconds instead of a shared staging environment that one agent corrupts for the next. He tells the story behind the company: turning down big-tech offers after his master's, a year of failed products built from his parents' house in Seattle, a one-way ticket to San Francisco that went viral, a year on a couch, and a Y Combinator acceptance on the eighth application. He also lays out his playbook for founders: sell before you build, post consistently because you cannot predict what resonates, raise only what you need, and keep equity for the people who make the company succeed.

Vikram Chennai
Vikram Chennai, Founder and CEO at Ardent.

Vikram Chennai is founder and CEO of Ardent, which clones a production Postgres database in under six seconds so coding agents can test their work in parallel without touching production.

Vikram turned down big-tech offers after his master's degree and spent a year building out of his childhood bedroom in Seattle, shipping four products nobody wanted before an AI data engineer found paying customers. He bought a one-way ticket to San Francisco with $10,000, slept on a couch for most of a year, and got into Y Combinator on his eighth application after pivoting to database sandboxes. Ardent raised about $2.2 million before YC and grows on inbound from X and LinkedIn.

“You have no idea what will go viral. If you keep not taking shots, you have a zero chance to figure out what it is.”

Vikram's core argument is that founders should sell before they build, because a product with real demand onboards customers on the first call instead of convincing them.

"The real parts of building a company are being terrified that everything is going to go completely wrong," Vikram explains. "And figuring out how to just take the next step."

The conversation

In this conversation: Jacqueline Samira (CEO, Howdy) and Vikram Chennai (Founder and CEO, Ardent).

Recorded in San Francisco for the CEO Studio interview series. Interview recorded on 07/30/26.

This interview was recorded by CEO Studio, a media brand owned by Howdy (howdy.com). Transcript lightly edited for clarity.

Vikram Chennai

Hi, my name is Vikram. I run a company called Ardent. We do Postgres cloning for your database in less than six seconds, so your coding agents can work in parallel on your database without any risk to production.

Jacqueline Samira

Can you explain that to a non-technical person? What does that mean to them?

Vikram Chennai

One of the biggest things we've seen with coding agents is that they can run in parallel. You can do 10, 20, 30, 100 things all at once. But the problem with something like a database is that it's really heavy. If you want each of those agents to work on the production state of your database all at once, you need a way of cloning it out. We've already done this with code, with Git and GitHub, so we're taking the same model and doing it on every Postgres database out there. What that results in is agents doing 100 tasks at once, working on the exact state of the world that all of your customers are interacting with, in parallel, solving problems at massive scale.

Jacqueline Samira

This seems to be a relatively new problem you're solving. At what point did you realize it was a problem and that you needed to go figure out a solution?

Vikram Chennai

I previously built an AI data engineer. That's where this whole company started. One of our biggest problems was that we would run multiple tasks in parallel, but you couldn't put them together, and you couldn't realistically test. So you would end up with a migration on your database that changes the state but doesn't actually work, and no one knows until it hits production. That's a recipe for outages. We realized we couldn't parallelize this. It didn't scale as a unit. We needed infrastructure that scaled, gave us the production state, and could scale independently of the production database without changing anything. Now we can unlock a world where agents can say, you want 100 different solves all at once? We can do that in parallel without touching the database.

Jacqueline Samira

Why couldn't somebody like GitHub do this?

Vikram Chennai

One of the hardest things about it is that data is heavy. That's the core of it. You're dealing with the laws of physics. GitHub is dealing with some number of files, and files are pretty small, so cloning a bunch of files is not that hard. For data, you have terabytes and terabytes. Even moving it is a challenge, which is why it's not so simple to say, I'll clone my database ten times. You have to build specialized infrastructure and do a bunch of tricks to get around the way data moves, just so that this is possible.

Jacqueline Samira

What are companies doing right now that aren't using your product? How are they getting around it?

Vikram Chennai

They don't. The traditional way has been staging environments. You clone a subset of the data, or you clone everything once. It takes anywhere from six to eight hours per terabyte, so companies with even slightly large data sets hit this wall. They're usually syncing it every night, something breaks, and they wait another day to do any work. So what people usually do first is point all their agents at staging, because they have their clone. But if they're all writing to the same place and testing in the same place, they clash. One agent's work overwrites another. You don't have a clean state for them to work on, and you don't know if any of the work actually worked. Agent one is corrupting the state of agent two. Then they move to pulling a subset of data. But a subset is not the entire data set. If you have an anomaly in production that you never pulled into your subset, it still exists. You fire off a migration, build a new feature, and it fails in production. Now you have an outage, and outages cost a lot of money. What we do instead is let you clone your database instantly, in less than six seconds. You don't pay hours and hours of tax. It's a managed service. All you have to do is tell your agent, solve this and solve that, and it will just do it.

Jacqueline Samira

So you're setting up a bunch of different sandboxes for all these agents. For the non-technical person, how does a sandbox differ from setting up different staging environments?

Vikram Chennai

There are two main things. Staging environments require you to move data, so you pay the tax of copying all of it. We've done a bunch of architecture tricks to eliminate that entirely. You can have petabytes of data and it clones in less than six seconds, and it's a full clone. You don't make any trade-offs about needing a subset. You get the entire thing, no matter what, and we can scale it out in parallel: hundreds, thousands, tens of thousands of clones. How? We do something called copy-on-write. Instead of moving the data, which runs into IOPS and bandwidth and all the laws of physics, we write pointers to the existing data. It's a pretty traditional computer science concept. The same data sits there, your normal database points to it, and if we create new compute and point that at the existing data using metadata, we move no data. Metadata is really light, so all we have to do is load metadata, and that makes it near instant.

Jacqueline Samira

When you're explaining your business to people with this pain point, do they get it instantly, or does it take some educational lift?

Vikram Chennai

It's a bit of both. A lot of our demand is inbound, so people show up ready to use the product. They know what it is and they've felt the pain. Our users are very technical. We usually work with a head of infra or someone like that, so they've already worked in this world. We have the architecture in our docs, so they read up and show up saying, I get it. The value prop is that they're already trying to run agents, they're already trying to solve this problem, and they usually have some hacky mess of scripts that didn't work or was too much maintenance. They see that we built exactly the solution to the problem. Oh, I can delete that problem overnight? Great, I'll do that.

Jacqueline Samira

What about fundraising? Were you able to find very technical investors who instantly got it after you launched?

Vikram Chennai

We were actually a little contrarian. Sandboxes were something people understood, but every investor thought a sandbox meant a traditional compute sandbox. A database sandbox is different. It solves the same agent parallelization problem, but it's a different part of the stack with really different requirements. A regular sandbox has a much smaller memory footprint than a database, because the database is humongous. If I tried to load all the database's data into memory, it would crash in a way a traditional sandbox never has to handle. So it was the super technical investors who were operators before, who had really experienced the problem, that got it. There were also a lot of posts on Twitter this year like, my agent deleted my database. That was a big driver: oh, I get it, that's the problem you're solving. People want to use agents on their database. It's a huge part of the stack, the whole data layer, and the question is how we make it possible for agents to work there. The investors who got that really got it. For the more general crowd, it wasn't quite consensus.

Jacqueline Samira

You said most of your new customers come from inbound. How are they finding you? And you mentioned you're growing double, triple, week over week.

Vikram Chennai

Most of them come from Twitter and LinkedIn. We post a lot, so people find us. For us, it's a game of: they will have the problem. It's not if, it's when. So we have to be there as the solution they already know about. We even streamlined our whole onboarding to just four CLI commands, because we know developers are going to hit this problem and reach for a solution, and they don't want to deal with setting this up and that up. They want their problem solved right now. So we made it super easy, and then they start using it. It's a usage-based product, so they get it into a workflow and see the magic. We talked to a user yesterday who booked time with us and spent the first ten minutes saying, this is magic. For the first time, he could kick off tasks in Linear, Linear kicks off a coding agent, and he could close his laptop, because he didn't need a local state of the database. It's just cloud production. It solved an entire issue end to end and pushed to production. He used to keep his laptop open in the car because he couldn't get the database piece. Now he just lets it go.

Jacqueline Samira

Is word of mouth starting to get out?

Vikram Chennai

Yes. We're getting to the point where people say, wow, this is really magic, I should tell other people. That person was actually a referral from one of our existing customers.

Jacqueline Samira

On X, did you have a following before? I talk to a lot of founders who are trying LinkedIn and X, and if they don't have an audience they feel like they're talking to no one. What did you have coming into this?

Vikram Chennai

I built a bit of a following, but I started from zero about a year and a half ago. I was in Seattle. I had a bunch of job offers coming out of my master's, and I said, nope, I'm going to start a company. I moved home for an entire year and built stuff out of my bedroom at my parents' house. I knew the only way I was going to get the real experience was to do the reps myself. I could take a nice offer at a big company, but I wouldn't learn what it's like, because the real parts of building a company are being terrified that everything is going to go completely wrong in every way, all at the same time, and figuring out how to take the next step and learn through that. I needed that experience. So I did that for about a year. I kept posting every now and then. It didn't really work.

Jacqueline Samira

For a year it didn't work? What were you posting?

Vikram Chennai

A little bit about the product. I wasn't super consistent, but it was enough to learn how to post. I built a bunch of other things, and all of it failed. All of it failed miserably. I learned very well what it looks like when you build something that literally no one wants. Then I got the idea for an AI data engineer, because my background is in model training. I knew that if I could improve data quality, everything downstream improves, especially model training. I talked to 50 or 60 people, ML engineers first, and got to data engineers.

Jacqueline Samira

Your parents are in Seattle. So you're in an area where you can meet people working on cool things and get feedback?

Vikram Chennai

Not really, actually. Seattle has been taken by big tech. One of the reasons I moved to San Francisco was that I wanted to be around people building from zero who got it. There's a community around that here that doesn't really exist in Seattle, which I wish it would. Every city has something it values, and what Seattle values is the white picket fence life. I work in tech, I make an obscene amount of money, I'm going to live this very nice life. It's not for everyone, and founders feel really out of place there. I felt really out of place. And no matter what, I want to win. I don't want to build a company just so I can say I'm building a company. I want the $20 billion exit minimum. So I bought a one-way ticket to San Francisco with $10,000. I liquidated all my investments and savings and slept on a couch for almost a year to make my company work. I wanted to be around the right community, and I wanted to give myself enough pressure that there were no other options. That's how the whole Twitter thing worked for me. I made a post and didn't think anything of it: hey, I bought a one-way ticket to San Francisco, with a picture of my ticket. It was just an accountability thing. If I put it out in the world, people will know, and I was really scared, so I couldn't back out. I put my phone down, and two hours later it started buzzing. Really big accounts in San Francisco had seen it and started reposting it, and it went from zero to a quarter million views overnight.

Jacqueline Samira

What were they saying when they reposted it?

Vikram Chennai

This is awesome. Yes, come to San Francisco. People were offering to help and DMing me. Investors DMed me too. My flight was three months out. The plan was to say it on the internet for accountability, expecting five likes, and spend three months getting ready. Then all this support happened and I said, nope, I'm doing it this weekend. I didn't even wait. That kick-started the whole Twitter thing. I'd never sold a thing in my life. I'd never really posted. I was figuring it out for the first time. From there I got enough reps to know what good looked like, to the point where if I want 100,000 or 200,000 views, I can probably get it within two weeks. You learn what works, what resonates, and you build an audience that resonates with what your account represents. Then you find the general narrative that makes sense and keep pushing on that.

Jacqueline Samira

You found alignment with an audience that resonates with your message. There are thousands of versions of our personalities that can find an audience, but it seems super aligned with who you are and what you're building.

Vikram Chennai

The number one thing people don't do well is they think they're not interesting, and they usually are. They hesitate because they think they need to be doing this really cool thing, or changing the world tomorrow, otherwise no one will care. There's some truth that really interesting things are really interesting. But most followings get built by just talking about what you're doing, because it's way more interesting than you think. And that's the biggest thing: you have no idea what will go viral. You have no idea what will do well. If you keep not taking shots, you have a zero chance to figure out what it is. I didn't expect that post to go viral. I thought I'd get five likes. My biggest post before that was 300 views. So keep posting consistently. You will eventually find something, because you're probably interesting enough, and then you figure out how to compound it.

Jacqueline Samira

This is an important conversation, because I talk to a lot of aspiring entrepreneurs in cushy corporate lives who say, I don't have a Twitter following. When you do sell for over $20 billion, people will look back and say, well, he already had an audience. They don't hear about the moment before, when you were at zero, trying things, and your biggest post got 300 views. Then you went accidentally viral and learned from it. That's how everybody grows, but they miss the zero to one journey on the audience side.

Vikram Chennai

It's not just me. A bunch of other people have grown a following. It used to feel like luck, and the more I did it, the less it felt like luck. Especially for socials, there are ways to engineer it. It's just really opaque, so you have to do enough reps that it makes a little more sense each time, until you feel like you have a degree of control over it. Even if it's volatile, you know that on average you'll get 100,000 views a month, because you've figured out enough to be at that level.

Jacqueline Samira

Do you approach your audience differently based on the medium? Is how you show up on LinkedIn different from X, or does the formula work across platforms?

Vikram Chennai

I think it works across platforms, but it depends. A lot of my followers are very supportive of what I'm doing, and that's a narrative you can tell on both platforms. Of course there are posts that went crazy on LinkedIn and flopped on Twitter, and the opposite. What I tend to do is post the same thing on both, because there's not much risk. It's different for each person, but for me it's hard to split, because I'd have to think of things I would never say elsewhere. I might package it a little differently for LinkedIn versus Twitter, but for the most part I'm saying the same things.

Jacqueline Samira

You're a solo founder with a very small team, growing like crazy, and you still find time to be on X and LinkedIn. If you were talking to new founders, how much of their time should be marketing versus building? What does that blend look like?

Vikram Chennai

I personally think you should market first, because if you don't have a market to sell into, it doesn't matter what you build. I've made this mistake before. I'm an engineer, so I've built products that literally no one wanted. That's where the four failed products came from. This was earlier in AI coding, 2023, so you couldn't magic a product out overnight. It took a while to build, I launched it, five people cared, and then they churned. What I should have done, and what I did for the AI data engineer, is build the smallest prototype and just sell it. If people are willing to pay me thousands a month, I found a real problem. My confidence then came from: all I have to do is make it work. That might be really hard, but if I make it work, I have something. Versus, I don't know if I have anything, and this work might all go down the drain. That's way more draining. So do that first, then build. For products with real demand, you should be able to sell before you build. Put something out there: a landing page, or ask the 20 closest relevant people. If someone pays for it, someone wanted it. You can refund their money and be good about it, but that's where I'd start. Then when it comes to juggling both, the company has to grow. They're both non-negotiable. So now it's about hiring the right people, and being very careful, because if you don't, you lose more than you gain. Our CTO is awesome and has taken a lot off my shoulders, where I was originally doing everything. That's the natural evolution: you parallelize the work to the right people so you can focus on what you need to.

Jacqueline Samira

I completely agree with getting buy-in first. Go talk to 40 or 50 people and get some traction before you spend time, money, and resources building the thing. The problem is building is so much easier now that people naturally want to build first and sell later. Doesn't it feel so much better when people want the thing you build?

Vikram Chennai

It just works. When we sell to people, we don't really have to convince them. It's them asking, does it do this thing? And we say, yes, it does exactly that. Great. We're effectively onboarding them on the first call rather than selling them. There's clear demand. Usage goes up every week. If that wasn't there, it would be really hard. A big mistake people make, even if they raise a lot of money, is scaling a team around something they're not confident in, without a clear demand signal. Then they have to raise the morale of all these people without the fundamentals that are supposed to do that: the number going up every week, a new person really liking it and telling everyone how great the product is. Those are the things that create the momentum that keeps a company alive.

Jacqueline Samira

I used to tell folks, if you have to sell something, your product is not good enough. The whole thing should be discovery. Find out what their problem is, and if your solution fixes it, share it. People overcomplicate so many things. Does it do this thing? Yes. Great. Let's go.

Vikram Chennai

That's it. Find a few people who will pay for the thing. People think of sales in a weird way, like, I'm trying to convince them to buy my thing. Really, they're happy to pay you to delete the problem for a certain price. You just have to find something people care about enough that the economics work for everyone. It's less about making so much money and more about making the wheel go so it can get bigger and serve more people.

Jacqueline Samira

Let's go back to pre-YC. Did you incubate this in YC, or was Ardent already a product out there?

Vikram Chennai

The AI data engineer was. I built that for a year before YC. Before YC we raised 2.2 million, just me, solo. I sold a bunch of ARR, delivered, built the AI data engineer, and did that for a year, all solo. That was particularly tough, and I cut my teeth on the hiring lessons: who to hire, who not to hire. You'd think with a decent Twitter presence you'd find people who are super interested, and it never works the way you think it will. But eventually we found the right people. Then around last December we pivoted. The big problem was I couldn't validate any of this work, and it turned out everyone probably had this problem too. We brought it into YC as an MVP. It worked, but only kind of. We had three customers paying us $100 a month, our earliest three people, who said, whoa, this is really cool, I want to use it. It would be buggy, and they would help: here's the branch ID that's failing, here are all these things, I want to help you make it better because I care about the product and it's really valuable to me. That's what we came into YC with, and then we ramped through it.

Jacqueline Samira

So you pivoted right before YC, and it's the same entity. Did you raise more money after YC?

Vikram Chennai

A little bit more. We decided to raise just what we need, because we had enough. By the time we got to YC it was just me, so we did not have a high burn. Our product is profitable on a margin basis; we don't lose money serving a customer. So it's a game of, we don't need infinite money. It's more economical to shoot for our Series A than to raise a ton now and go through that whole process. Get a little more capital, find the right people, keep going. The other thing I care about personally is that I'm very greedy for equity when it comes to taking on capital, because the people who make the company succeed are investors and the right people, but also my employees. I want to reward them as well as possible, and as a solo founder I can give people outsized equity so their work is rewarded. If I did a big round, there would be less for the people who deserve it. And if we're going to raise a Series A anyway and we have enough, it's just capital for capital's sake at a price that's not optimal. Why would you do that?

Jacqueline Samira

A lot of founders chase the size of the raise without seeing what it means to their cap table and the employee pool. So what encouraged you to apply to YC when you'd already raised 2.2 million?

Vikram Chennai

The way I look at selling equity, because that's what you're doing, is: does it improve my odds of success? Capital can, network can, community can. A few friends who had done YC recommended I apply again. Before that, I'd applied seven times and they said no, for a few different products and the AI data engineer. This was the eighth. I didn't need the capital, but it's a great community with great people, and my friends said, you're trying to get to 20 billion. Does it improve your odds of success? Probably. And it can't hurt to apply, you've done this application enough times. So I threw in an application, got an in-person interview four days later, and then got an email from Aaron, my group partner. Funnily enough, it went to spam, so I freaked out for an entire day thinking no one had responded. I got the call a day later: do you want to do YC? The decision was, does giving someone this equity improve my odds? If my target is that high, I need every advantage. Money by itself doesn't do enough. I need the leverage of the network, of finding early customers, of the brand. All these things compound. I need the best engineers in the world. My CTO has 20 years of experience as a data engineer. A lot of the bet was on me as a founder, but a lot of it was also, you've accumulated the resources for us to win. Now we have what we need to get the best people and compound the company, so that when anyone looks at solutions in the space, they say, that's the best product by a mile, and it's built by great people.

Jacqueline Samira

Very similar reasoning for us. We were already doing 1.8 million in revenue when we applied to YC, and giving up a sizable chunk for a low amount of money, everyone asked why. The compounding effect of being part of YC. We were also building something new in a space with a lot of historical stigma, and saying we got into YC made everyone pause and ask, wait, what are you doing again? It was the best decision we ever made. Your story is incredible, and I'm so glad we get to share it, because what people don't see when they see the success is the journey that got you there. You slept on a couch for a year. You went from zero to a following in a year and a half. You applied to YC eight times. I can't wait for you to be more than a $20 billion company, because you definitely will be.

Vikram Chennai

I appreciate it.

Jacqueline Samira

Thanks for coming.

GET INVOLVED

Be part of the
conversation.

Whether you’re a CEO or founder who wants to be featured or a company interested in sponsoring, there’s a place for you here.