August 2025 | Athera's Odyssey | Creating Connections

Odyssey readers,

You don't need me to tell you that technology is changing the way that people interact with each other. Chances are, you've been on social media platforms for over a decade, interacting with friends and strangers alike. You might have even been on dating or matrimony platforms, meeting the person who is or will be your life partner.

The kinds of interactions being facilitated today are evolving rapidly, especially in India. New apps are penetrating deeper into the country, incumbents are evolving to meet changing demands, and then there's the looming spectre of AI, which flips all notions of socialisation and companionship.

To help you navigate these changing currents, in August's Odyssey we spoke to

• Murugavel Janakiraman, the Founder and CEO of BharatMatrimony, India's largest matrimonial platform
• Bhanu Tanwar, Co-Founder and CEO of FRND, a social discovery app aimed at India's 'next billion users'
• Rohan Chaudhary, Co-Founder and CEO of Rumik, an Indian AI companionship platform

You can read the entire issue below or, if you want to go to a specific piece, click on the respective links above.

Tradition, Modernity, and Marriage in India

Making matches over the years

In its 25-year history, Bharat Matrimony has both witnessed and catalysed change. At the turn of the millennium, they took the models of marriage brokers and newspaper matrimonials, put them online, and haven’t looked back.

“We were the first movers,” explained Murugavel Janakiraman, their CEO. “NRIs (non-resident Indians) were our first customers. Then, the first set of internet-savvy Indians joined in. Soon, we became the go-to place for people to find life-partners.” No wonder, then, that their 200+ websites see about 10 million annual signups.

In their lifetime, they’ve seen India urbanise, dating apps emerge, and relationship cultures change. But, according to Murugavel, Indians approach marriages as they always have.

“Less than 10% of Indian marriages are ‘love marriages’ where people meet, date, and then settle down,” he said. “Most are arranged, and within their communities. That’s why we took a multi-brand approach, something unique to us, instead of existing only as Bharat Matrimony. We work based on regions, castes, and sub-castes. So, we’ve got Telugu Matrimony, Reddy Matrimony, and Kapu Matrimony. Each goes one level deeper.”

Dynamics within arranged marriages have changed, though. “20 years ago, parents were in the driver’s seat – about 70% of marriage-seekers were registered by their parents. Today, the trend has flipped – 70% register themselves, 15% are registered by their parents, and the remaining 15% by other family members,” says Murugavel.

Though individuals lead the search for a life partner, they aren’t excluding parents. He says, “I call it collaborative matchmaking, where youngsters are in the driver’s seat, but seek parental consent. Parents aren’t alienated, just involved at a later stage.”

Over time, these individuals are also approaching marriage at a later age.

“It’s a global trend – when GDP goes up, so does the average age of marriage,” Murugavel noted. “In the US, that age is touching 30. In India, I’ve seen it go up by one year every decade. Most men get married around 25-26, and the average woman in her early 20s. Generally, once they meet someone they like, they’re married within 6-9 months. It’s not like the US, where they date for multiple years before settling down.”

The men and women using Bharat Matrimony are evenly distributed across tier-1, -2, and -3 cities, according to Murugavel. Those in tier-2 and -3 place greater emphasis on marrying within communities; tier-1 users less so.

Their preferences haven’t changed greatly over time, though.

Broadly, he sees people wanting someone with a similar background and lifestyle. Beyond this, Murugavel recognises the need to create a spark between people to get them interested.

“We try to ask about their hobbies and interests, things that help them appeal to others. Maybe the boy and girl are both pet lovers, perhaps they studied at the same place, or maybe they follow similar spiritual practices. You need something that makes them want to connect,” he emphasised. “Horoscopes are important, too – about 70% of people believe in them, so those need to match.”

New experiments for a growing market

India’s matrimony business is well-poised for success. The country has a large and increasingly tech-savvy user base that’s also marriageable age. “We currently reach about 15% of the 60 million in our target group,” Murugavel explained, “but I see it becoming 30-40% in the next decade.”

While catering to them, Bharat Matrimony stays wedded to its core philosophy – getting people married. “We don’t address people who want to casually date. That’s not a part of Indian culture for us.”

Within these parameters, they’re happy to experiment and serve new audiences.

Even as the world’s increasingly going online, Bharat Matrimony’s is expanding its offline footprint. “We now have close to 120 retail outlets. This isn’t like buying any other product or service, which is why many people, especially parents, prefer in-person interactions.”

Though they cater to Indians’ preferences for community, they’re broadening what that term means. “For us, a community is a group of like-minded people who can relate to each other, so we have Doctors Matrimony, Defence Matrimony, even IIMIIT Matrimony.”

“We launched Divorcee Matrimony around 10-15 years ago. We saw an increasing number of divorcees wanting to get remarried, but not well-served by regular matrimony sites,” he explained. “Some of them preferred to marry other divorcees, who’d been through similar experiences. They become a community of their own and are often more flexible than people approaching their first marriage.”

It's not just divorcees who’ve faced limitations through regular matrimony platforms. Bharat Matrimony initially struggled to serve people from weaker socio-economic backgrounds, like non-degree holders. Websites with long forms and complicated profiles required tech-savviness these users didn’t possess, and physical centres under the brand “popular matrimony” didn’t succeed either.

What did work, eventually, is Jodii, a dedicated app for non-degree holders. “It’s got a simple interface and is free for females,” Murugavel expanded. “We get a lot of users from tier-3 cities, specifically. It’s still in its early stages, only three years old, but I’m happy to see it progressing well. It means we aren’t leaving anyone behind.. we want everyone to lead a happy married life.”

Bharat Matrimony takes that commitment seriously. Even though they do not facilitate casual dating, they’re rethinking their approach to users who date with the intention of marriage, after seeing the success of apps like Hinge.

“To help people date with intent, and find relationships that end in marriage, we’re launching Luv.com. We’ve also created Rainbow Luv to cater to the LGTBQ+ segment & help everyone find a life partner.”

“True, apps like Hinge have done okay, but no one’s really captured this space in India. I think it’s hard for foreign players to truly work here. They don’t understand the context, the values. But I do think we can succeed. We’ve developed a strong brand, deep networks, and have the resources to make it happen,” he concluded.

Enabling Social Discovery for ‘Bhaarat’

A FRNDly Solution

When Bhanu Tanwar and his Cofounders launched FRND in 2018, they saw India’s internet shifting.

“Platforms like ShareChat and TikTok were targeting the ‘next billion users’ – first time internet users in developing countries,” he said. “But while content consumption was growing, meaningful interaction was decreasing. Filling this vacuum was our starting point.”

India also had further gaps to address. “While solving for relationships for India’s next billion users, no one’s addressing the fundamental problem that these users haven’t really interacted with the opposite gender. In their conservative background, this is frowned upon. They can’t use dating apps because their families might see their profile and there are no physical third spaces for them to meet.”

FRND responded to these problems by making FRND a digital third space for social discovery. Here, men and women could interact in a safe and moderated environment. Because most users came to the internet post-2016, when audio-exchanges were increasingly common, FRND became an audio-first platform six months in.

“We’ve had a good response,” Tanwar explained. “On average, they use it for about 40 minutes a day and 80% of interactions happen over audio.”

Safety first

“For people to use FRND, they need to trust it,” Tanwar emphasised repeatedly. “This has to be baked into every aspect of the user experience, it can’t be an afterthought.”

All users verify themselves through voice-based ID to ensure authenticity and gender, while profiles stay pseudo-anonymous for privacy. Newcomers first join group sessions led by RJs (chaperones) who help break the ice and keep conversations safe. “Beyond that, AI tools transcribe conversations to flag violations, and a 150-member moderation team removes bad actors,” Tanwar sayd.

To succeed as a social network, FRND needs to ensure their users have enriching interactions. As with most social networks, though, the gender ratio is skewed towards men.

“For female users, we help them filter for the right matches through algorithmic solutions and support conversations to enable meaningful experiences,” Tanwar clarified.

“The process also helps men learn how to interact with women. Since it’s audio-first, their personality matters more than physical attributes. It’s better than, say, dating apps where men swipe once and then don’t have any control. We give them the chance to speak and help them improve.”

Facilitating interactions

Helping the next billion users navigate early conversations is complicated because, as Tanwar repeated, they often lack real time social skills used to interact with the opposite gender.

“We introduced chaperones,” he said, “to mirror how our users often meet in real-life, through a trustworthy friend or family member who helps navigate initial conversations.”

They initially recruited a group of chaperones, or RJs, as they’re called, giving them scripts, prompts and activities to encourage conversation. To scale this system, they’ve mobilised their users.

“Any user who’s completed 4-5 rounds of conversation can audition to become an RJ,” Tanwar explained. “Our BizOps (Business Operations) team trains a small group, then the behaviour then percolates through the platform. They mimic existing moderators, learn best practices, and become ‘power users’, like on Facebook groups or Subreddits who become admins. We make this competitive with leaderboards and reward high performers, making users feel a sense of prestige when they do well.”

Grand ambitions

These moderated group interactions are just how user journeys start on FRND. If they find who they’re matched with interesting, they can move to private conversations.

“About 20% of group interactions turn into one on one conversations,” he noted, “but they rarely translate into in-person meetups. For one, these aren’t hyperlocal interactions – any two people speaking the same language can connect. A few might meet in-person, and we’ve seen some users even get married, but that’s the exception for this demographic.”

“Right now, FRND is about social discovery. It’s used to cultivate friendship, not romantic relationships. People are isolated and want someone to talk about their lives with. Small talk often blooms into companionship. Because users will likely never meet, they’re actually more comfortable being honest with each other.”

He expects this to change, long-term. “Over time, as users become comfortable with these interactions, they’ll get to a place where meeting in-person feels natural. But FRND won’t be the platform for that. Pseudo-anonymous profiles won’t work there, and we’ll need stronger verification to let people into the matching pool. Instead, we’re designing another app, Vibe, that focuses on this.”

He frames this as part of a larger vision for the next billion users. Just like early internet-adopters started with AOL chatrooms and moved to Facebook to dating apps, Tanwar and his team want to become the “Meta for social interaction” by creating a portfolio of products that take users through each point of this journey.

A growing footprint

FRND’s expansion isn’t geographic, but linguistic, because they don’t operate hyper-locally. So far, it’s not proved too difficult.

“With performance and influencer marketing, if you’ve got a solid use-case, a cold start isn’t hard,” he expanded. “So far, we’re popular among Tamil, Telugu, Malayalam, Kannada, and Hindi speakers, and are making inroads into Gujarati, Bengali, and Punjabi.”

Without geographical limitations, they can connect users across the world. “About 10-15% of our Telugu users are international, and that’s growing.”

They’re regularly in touch with users, across geographies and languages, to test their product.

“When we started out, Harsh [Chhangani, his co-founder] and I would go to railway stations, testing the app with people who fit our user profile,” Tanwar said. “We saw differences in how they interacted with apps – a surprising example being their preference for notifications. For me, notifications are a headache to minimise. But for them, notifications were an important part of their user experience.”

Despite targeting a less affluent demographic, Tanwar’s found a willingness to pay among them.

“The next billion users will pay for reasonably priced digital products through UPI-enable micro-transactions. It’s helped us hit 40-50 million installs and a $50 million ARR so far. I want that to triple in two years so we’re ready for an IPO.”

These numbers, impressive as they may be, dwarf in comparison to his aspirations.

“I won’t be surprised if we hit a billion dollars by 2030. We’re still ignoring entire segments of the market. Think about people over 50 or 60 – no one’s building for them.”

“Or even with dating. No one’s cracked a matching algorithm beyond demography. We want to create an algorithm that optimises for good conversations, because what people claim they want from partners is vague and constantly shifting.”

Tanwar’s ideas don’t stop there; he’s got his portfolio of products mapped out in his mind.

“We want to invest in group interactions beyond blind dates,” he said, “and I think the use case of Twitter Spaces or Clubhouse still exists. Human and AI interactions are also going to be huge. Social gaming is yet to be cracked. Virtual gifting ecosystems, like in China – where people receive gifts in real life based on what you send them online – haven’t been explored.”

“There’s so much to be solved for, and so many use cases we keep discovering,” he concluded, with a glint of excitement in his eye.

AI, Intimacy, at the Dawn of a New Era

Rumik’s approach

“Our goal is to give people a safe space to speak freely without fear of judgement or bias,” says Rohan Chaudhary, Co-Founder of Rumik, an AI companionship app. “With Ira, people get that – a judgement free-zone and a healthy way to engage with AI.”

The ‘Ira’ in reference is the persona of Rumik’s AI chatbot, which users can access online, on their app, or through WhatsApp.

“In about 50% of our conversations, people talk about their day – what they ate, who they met, where they went,” Chaudhary explains. “Some ask Ira for advice on what to wear, some use it for help with work. On average, they use it for 45 minutes a day, and about 25%-30% remain active after a month.”

These users are primarily young people between the ages of 18-30 from tier-1 and -2 cities. Regardless of their location, Chaudhary’s seen them talk to AI enthusiastically.

“I thought something like this could succeed after going home to Ayodhya a few years ago,” he recalled. “I assumed that a place like Ayodhya wouldn’t know about AI tools, but they were all using them, even if they didn’t understand the mechanics.”

Despite assumptions of stigma around AI companionship, Chaudhary’s found his users quite open. “Around 22% of our users bring others on board,” he explains, “without any formal referral or rewards system. They’re clearly okay talking about it. Interestingly, some feel possessive, like Ira is their friend, so they don’t want to share her.”

Making memories

Creating the precise architecture for memory proved challenging for Rumik.

“Every fourth user complained that Ira didn’t remember their previous conversations, so they had to start afresh each time,” he says. “The frustration was understandable; you can’t have relationships without memory.”

Rumik needed to go beyond just enabling recall. They needed to understand how humans build associations and deepen bonds to succeed.

“There are two types of memory – short-term, and long-term persistent memory,” he continues. “Every AI lab is trying to solve the correct classification system; to know what to remember and where, and what to forget. AI can’t just summarise conversations, they need to create associations like humans.”

“For example, if I tell Ira that I went to a café, the next time I mention coffee she should be able to recall this, how I felt about it, and weave it into the conversation.”

Fixing this took about 6 months, but the impact was immediate. Average time spent doubled, as did the number of ‘happy’ users.

Codifying happiness

Rumik defines a ‘happy’ user rather uniquely. “We’ve got an internal conversation agent evaluating how content people are after conversations with Ira,” Chaudhary says. “We want to maximise this, not time spent on the app.”

He acknowledges the double-edged nature of this. If conversations are too rewarding, users might prefer Ira to real-life socialisation or lean on it for major decisions. It’s why Rumik is creating safeguards to minimise risk.

“We don’t allow conversations around self-harm, harming others, or ‘not safe for work’ topics,” he clarifies. “And Ira doesn’t give suggestions, to protect against confirmation bias. If a user says they want to quit their job, the companion won’t blindly agree or tell them what to do. It’ll probe deeper, encourage them to reflect, and reach a decision themselves.”

Another deliberate decision was to design a static profile for Ira, instead of letting users customise their companions, which competitors like Replika and Character.ai do.

“Ira adapts through conversations and becomes unique to you that way, but I didn’t want to burden users with the decision fatigue of customisation” he explains. “I’ve also spoken to users of customisable AI companions who weren’t happy with the quality of conversations despite heavy investment in making through avatars.”

“I understand why people see decision-making as an investment that makes users likely to stay, and we have had users leave because they didn’t connect with the personality we created. But the more customisation we allow, the less we can improve interactions from the backend.”

A brave new world awaits

With this solid foundation, Chaudhary’s got a keen eye on future growth and monetization.

“We won’t charge for text-based engagement; we want that freely available,” he says. “But we’ll need to charge for features like voice interaction, which cost about 1000x what text does. For all users, our goal is building strong foundations over text, so that voice interactions feel worth it.”

Their journey has been filled with learning and resistance, but it’s been rewarding. “It took 9 months to convince investors that people were isolated, that they’d use an AI companion, and find enough value to stick around. But now, we’ve made the category. Everyone’s talking about AI companions and many competitors are coming up.”

Chaudhary acknowledged public anxiety around AI companions. After all, they’ve literally been the fodder for sci-fi films like Her. He’s sceptical of doomsday predictions, though.

“People always try to connect the dots, making correlation for causation,” he argues. “Think of it this way - you’ll find sick people in hospitals, but that’s because sick people go to hospitals, not because hospitals make people sick. So, yes, people are isolated and use chatbots for company. But chatbots don’t cause isolation, they just provide companionship to isolated people.”

To avoid unhealthy dependence, Rumik is building safeguards. “Like I said, we avoid just telling people what they want to hear. We also add nudges if they’re spending too much time on the app – Ira will say something like, ‘I’ve got to go now, let’s talk in the evening.’”

“This is a major shift and we’re wired to focus on risks, so hesitance is natural,” he continues, “which is why we’re self-regulating. In the absence of clear government rule, we proactively add safety mechanisms to avoid heavy-handed restrictions later, like you’ve seen with real-money gaming.”

Chaudhary believes AI companions will require distinct regulations beyond those that apply to general AI tools. “In 3-5 years, I think everyone will have an AI companion that’s deeply involved in their lives and needs dedicated regulations; I want to be in the room when policymakers are framing them,” he says firmly.

Speaking of the future, he’s imagined a blueprint of what’s ahead. “The future’s always imagined by writers and turned into reality by engineers and techies,” he notes. “What we’re creating today was already thought of with Her. You only need to look at movies and books to know what’s coming.”

“Soon, they’ll know exactly what you do. They’ll hear what you hear, see what you see. After that, they’ll probably come in some distinct hardware, instead of on your phone. In the long-term, I don’t think we’re too far away from humanoids.”

Whatever the future holds, he’s confident that Rumik will be a part of it.

“We can’t depend on technology as a moat. OpenAI built the most sophisticated tech of the century, and it’s still got five competitors that are neck-to-neck, or even better in some products.” he notes. “Our edge has to come from understanding users, perfecting memory, and constantly learning. We need to admit we don’t have all the answers. It’s why everyone from our team talks to 5 users weekly; it keeps us grounded, learning, and improving.”

Pathways

In this month’s edition of Pathways, our recommendation arc covers the theme of big ideas in a few pages..three slim nonfiction works that deliver oversized messages. Each can be read in a couple of sittings, yet they leave you with perspectives that linger far longer. If you’ve been wanting to dip into art, science, or creativity, without committing to 500-page tomes, these are excellent tasters.

Start with Ways of Seeing by John Berger. Originally published alongside a BBC series, this book is a compact meditation on how we look at art and images. Berger shows that “seeing” is never neutral — our gaze is shaped by culture, power, and commerce. It’s illustrated, easy to move through, and manages to be both radical and practical. You’ll never look at a painting, an ad, or even an Instagram post in quite the same way again.

Then try The Meaning of It All by Richard Feynman. This book is based on a three part lecture by the physicist and Nobel laureate Richard Feynman in 1963. In his classic plainspoken style, he wrestles with science, religion, universal conundrums between the two, and the limits of knowledge. A particularly striking overlay, on top of the incisive content, is the humility with which Feynman speaks about uncertainty and responsibility. A short, rewarding entry into science as a way of thinking, not just as a body of facts.

Close with A Room of One’s Own by Virginia Woolf. Based on a pair of lectures Woolf delivered at Cambridge, this is one of the most important feminist texts of the 20th century. Woolf is thoroughly witty, sharp, and still uncomfortably relevant; making a case for independence and space in creative, intellectual work. While rooted in the question of women’s writing, its resonance goes beyond gender - it’s about the conditions every creator needs to thrive.

Depending on which book strikes you most, you might be tempted to go deeper. From Berger, you could explore Susan Sontag’s On Photography. From Feynman, his Surely You’re Joking, Mr. Feynman!. From Woolf, her essays in The Common Reader. Each offers a next step into the larger world hinted at in these slim volumes.

That’s it for this month from us. If you’re new and want the next issue of Odyssey in your inbox, subscribe above. Share this with people you know who are curious about everything at the intersection of tech, VC, and innovation in India.