Subhash Kalluri Built FreJun's Voice Infrastructure Without a Single VC Round

q01Take us back. What happened before FreJun even existed?

Subhash Kalluri: Okay, actually, I'll just give a very brief background. Before founding FreJun, I was working in the US, and I thought maybe it's time for me to try something out. Without a plan or anything, I just resigned one fine day and moved back to India. Like any over-positive and over-enthusiastic first-time entrepreneur, I was really optimistic that it was going to work out.

The first couple of years were really tough. I tried to build a couple of products, and I'll be honest, I failed. Then I started building a different product called Minutes, basically whatever now we have like Fireflies. We built a similar product to that just before COVID, and unfortunately, there was a huge demand for it, but the biggest challenge was we did not have AI this advanced back then. We are like six or seven years back. The models were very primitive, and there were a lot of mistakes in the transcription.

I met a couple of VCs in Bangalore to raise funds, and the biggest challenge, from their point of view they're right, was I did not have a tech co-founder. I had a tech team, just out of college, and we could scrap up a product that actually works. But the transcription was bad, so the meeting summary was going to be bad. And I could not raise money. So I had a product with no customers.

q02By then you'd already spent three-plus years on this. What made you keep going instead of walking away?

Subhash Kalluri: By then, I had already spent three-plus years on this entrepreneurial journey, and I think any logical person would have said, okay fine, I think it's time to wrap it up and go back to the workforce. But I felt, I mean it's not a logical decision and I would not suggest everyone to do it, I just said okay fine, I failed, but I want to try one more time, and then moved into this product.

Initially when we built the product, we never had an idea of building calling infrastructure or whatever we have now. That entire journey started with a simple product for recruiters to make calls and log the calls. So when our first version came out, what we realized is the revenue we are making, we are giving like 40 or 50% of revenue to telephony infrastructure providers. When I saw the bill, I thought it doesn't make sense. Why would we pay this much amount? I thought okay, let's see if we can figure out to build our own infrastructure. And we figured out a way to build it.

q03Was bootstrapping always the plan, or did it become the plan?

Subhash Kalluri: Raising money was the plan. The plan was always to raise some money. Unfortunately, since I could not raise money, I understood that this is a game I would hate to play, raising the fund, it's not for me. After sitting with a couple of investors, I could see myself attending a lot of meetings, and I thought maybe this is not something I'm going to succeed at.

I thought, what can I do within the limited resources I had back then to build a product and build a business? So the entire thought process changed from how can I forget about funding, to how can I get a customer, get paid by the customer, and actually build a business. Even now we are completely bootstrapped, and we raised only one round, like 10 lakhs from one of our first customers. He said he wanted to invest, and we said okay, let's do it. Apart from that, we never had any external investors.

q04Voice AI agents are changing what customers expect. What has shifted for FreJun?

Subhash Kalluri: Yes, oh a lot. The entire base of customers changed. Even the ICP, the Ideal Customer Profile, is a completely new customer set. Before, our primary ICP used to be managers looking to implement processes, but now our customers are CTOs, VPs of Product, senior management, and the end users are developers. Overnight we had to move from selling business-to-business to selling business-to-developers. Now we're competing like any other product where the end users are developers, similar to AWS. Different market sets, different ICPs. End of the day, what matters is are we building a good product for the ICP or not.

q05So how does FreJun actually fit into the voice AI stack?

Subhash Kalluri: So what we fit in is simply, we are the carriers of the voice agents. If a voice agent has to have conversations with people, for example a bank, when you call a support number and someone deployed a voice agent, the voice agent will be interacting with the bank's customer. The entire part of handling the call, the conversation, we are the infrastructure. Internally, what I say to my team is we are like the roads. We build the transportation for the AI agents to function. We are the layer on which agents are deployed, and we are like toll booths; based on usage, you're going to pay us a fee.

q06You've said the fancy stuff keeps changing but the infrastructure remains constant. Walk me through that thinking.

Subhash Kalluri: If you look at some of our competitors, they're doing multiple things, trying to do everything. That's entirely their prerogative because they have resources. Unfortunately, for us, as a bootstrapped company, we don't have the luxury of trying out multiple products at the same time. One product has to make revenue that can fund a second one, and the second one has to work.

My thesis is simple. Probabilistically, the agents will change, the LLMs and conversations might change, but the carrier part will always remain the same. So I thought, instead of us building voice agents, why can't we support something like CaaS, Communications as a Service, where we handle the entire infrastructure for someone to deploy voice agents, and let the customers grow on us. To simply explain, when there's a California Gold Rush, you don't want to go and dig for the gold. The goal is to build a lot of shovels so you can help everyone who is digging.

q07Is business communication heading toward voice-first, or is it going to stay channel-agnostic?

Subhash Kalluri: It entirely depends on the use case and which market we're talking about. Let's take India, where our parents' generation, 60, 65 plus, their smartphone adoption is very minimal. They use it for the phone and WhatsApp, maybe now a little YouTube and Instagram. But if you ask someone like that to have a conversation with an agent on WhatsApp, do you think they're going to type? I don't think so. They'd prefer to talk with the agent on a voice call. I come from an agricultural background, so I still see older people using old feature phones. Just imagine they can call a number and have a conversation with the agent that does everything a mobile app can do. That's an interesting use case for banks.

On the other hand, take booking a flight with Gen Z, they prefer chat. So it depends on the industry and the use case. I think email, chat, voice, and messaging are all going to exist together. It comes down to two things: how effectively are you engaging with your customer, and after deploying all this, are you capturing that information in a way that fits your workflow.

q08You've stayed fully remote while everyone else called people back to the office. What made you commit to that?

Subhash Kalluri: That's a fantastic question. It was not out of choice, it was out of necessity. We stayed remote because we did not have money back then to invite everyone back to the office, that's the truth. It's also much more cost-effective because we don't have any overhead costs. And once you're remote, you can hire across India, and if you're lucky across the globe, because your applicant pool increases multiple-fold. You don't have to pay the same competing salaries for someone living in a major city. For a role, if I have someone in Bombay, I have to shell out at least 30 to 40% more.

But it comes with its own drawbacks. Not everyone is suitable for remote, so there are certain personality traits required to be effective in remote working conditions, and we look for those when hiring. Our entire company runs on Notion, literally. Whenever we have a call, we take down automatic notes and log them in Notion. The best way to describe how we work is asynchronous. The founding team is still with FreJun, and since we have that personal rapport, it makes it easier to work efficiently. But if I build another company, would I go remote? I'm not sure. You have to identify the drawbacks and build guardrails so they don't impact efficiency.

q09What qualities do you actually look for when hiring someone for a remote role?

Subhash Kalluri: Firstly is the maturity level of the individual. This is subjective, my personal opinion, but I look for how responsible the other person is, and most importantly, the temperament. The composure matters a lot because it's very easy, the human brain works when you're in survival mode, you always think negatively. So first is discipline. We look for at least one or two weeks to see whether this person is going to fit. Showing up on time, communication, are they over-communicating by asking questions, or are they just thinking this is a work-from-home situation and I can relax and take things easily.

I think mistakes happen when you don't observe these things after the individual joins. For example, we have company-wide calls, are they joining and participating? Even if they don't participate, are they joining? It says a lot about how serious they are about the work.

q10With a team spread across geographies, how do you actually manage sales coverage?

Subhash Kalluri: I'd like to take this occasion to give credit to our sales team, they work really hard. They work Indian Standard Time and European Standard Time, basically almost 12 hours a day, trying to work with the customer in their own time zone. They start their day a little late, but by the time they go to bed it's already 2 in the morning. If you're covering from morning 9 till 2am, you can cover all the way from Australia to the US.

We used to have two sets of sales teams working in shifts, but it wasn't really effective. One of the good things about FreJun is we don't have any fixed work hours and no official holiday policy. As long as you get the job done, we're fine. You can start your day at 10, you can start at 11, but there are parameters we track. If you're hitting those numbers, you're good.

q11Rapid fire round. First thing that comes to mind, doesn't need to be correct. One tool your remote team cannot live without?

Subhash Kalluri: Slack.

q12One habit that has helped you as a founder?

Subhash Kalluri: 20 minutes before you start your day, just understand what is a priority for the day and focus on it.

q13One mistake early founders should not be afraid of?

Subhash Kalluri: You're always wrong, and be shameless about being wrong.

q14One AI product you use every day?

Subhash Kalluri: Claude.

q15One underrated skill in B2B sales?

Subhash Kalluri: Replying on time.

q16One thing most founders overthink which they shouldn't?

Subhash Kalluri: Competition.

q17One thing most founders underthink but should concentrate on more?

Subhash Kalluri: Marketing.

q18Bootstrapped or funded, which founder sleeps better?

Subhash Kalluri: It depends. There is no bootstrapped or funded, it depends on what you're building, to whom you're building, and who is your ICP.

Final Thoughts: Building Roads, Not Chasing Gold

The bottom line: FreJun's story isn't a fundraising story, and that's exactly the point. Subhash couldn't raise money for Minutes, couldn't raise money for FreJun's early version either, and ended up building a company funded by customers and one 10 lakh check from someone who believed in it. The forced constraint of bootstrapping is what pushed him to own his own telephony infrastructure instead of bleeding 40 to 50% of revenue to a vendor.

The bigger bet is the infrastructure thesis itself. While competitors build voice agents that ride on top of someone else's rails, FreJun bet on being the rails. Agents, LLMs, and conversational interfaces will keep evolving, sometimes overnight, as Subhash's own ICP shift from HR managers to CTOs shows. But the layer that actually carries the call, the roads underneath the agents, is the part he's convinced stays constant.

For founders watching the AI wave and wondering where to build, Subhash's answer isn't glamorous: find the boring, constant layer underneath the exciting thing everyone else is chasing, and build shovels for the gold rush instead of digging for gold yourself.

q19How can people connect with you and learn more about FreJun?

Subhash Kalluri: You can visit FreJun at frejun.com. If your team depends on AI agents for sales, hiring, support, or operations, we already have the road, you just have to get your car and start driving.