As evening descends, it’s an all too familiar sight in Ranchi, Jamshedpur, Gwalior or Guwahati– Tier II and Tier III cities of India.
Groups of young men and women, often the college going crowd, huddled in circles, outside Chai Tapris (tea joints), fixated on their cellphones as they play their favorite video games, over tea and loud banter.
India accounts for world’s second largest mobile gaming market by downloads with an estimated 591 million active gamers in FY24, and 66% of them come from non-metro cities.
The intense popularity of mobile gaming, propelled by pandemic induced lockdown and comparatively cheaper internet rates has market pundits projecting $3.2 billion market size of India’s gaming industry by 2030.
The vibrant gaming community often needs a stage to connect and this is where social gaming platforms come into picture, allowing creators to connect with their fans in the bustling “voice rooms” and monetize their content.
However, with huge influx of gamers, the challenge to prevent incidents of cyberbullying, harassment, doxing and sexual abuse has only risen in the past few years. Gaming platforms consistently find it difficult to simultaneously monitor hundreds, if not thousands of live voice rooms in peak hours.
It is here that Artificial Intelligence (AI) comes into play as platforms integrate this advanced technology for not just AI voice moderation but also help gamers enhance their skills and keep hosting engaged communities at micro levels.
Ask 31-year-old Parth Chadha, the CEO of STAN Gaming, the importance of AI in his company, and he puts it bluntly that it would be impossible to run a gaming community platform with tens of millions of gamers without the advanced tech.
“If you removed AI from STAN, the product would not get slower…. it would stop existing at this scale. That is the test for whether something is core or bolted on,” said Chadha, in an exclusive interview with AI FrontPage, when he was asked whether AI sits at the core of STAN rather than being bolted upon.
Founded in 2022, STAN gaming is headquartered in Bengaluru city, with over 50 million app downloads and an average of 60,000 live clubs a day. While the numbers sound fancy, it could well appear as a daunting task to manage all gamers at once, given the diversity of languages and dialects in India.
AI voice moderation itself is not foolproof as multiple reports point out to challenges associated with voice speed, tone, and real-world chat that often comes with audio disturbances. Moreover, AI models have time and again failed to distinguish between harmless banter and actual threats leading to reactive decisions rather than preventive ones on these platforms.
In this exclusive interview, Chadha explains how AI has dropped the barrier for gamers in India especially those belonging to non-metro cities and how the technology still has a long way to go before it can independently monitor voice sessions.
Question: For someone outside the company, how would you explain the role AI actually plays day to day in making STAN Gaming platform work?
Parth Chadha: The simplest way to explain it: AI is what lets a small team run a platform for tens of millions of people who speak hundreds of languages, on phones that cost a few thousand rupees. It is not one feature. It sits underneath three things. First, safety. Every voice room and chat is moderated by AI in real time, because no human team could listen to 60,000 live clubs a day. Second, discovery. AI decides which creator’s room to put in front of you right now, based on what is actually happening in that room, not on who was popular last year. Third, creator support. After a stream, a creator gets AI feedback on what worked and what lost the room.
Question: India is an extraordinarily multilingual market, and you’re running this across hundreds of languages and budget Android devices. How does AI help you keep communities safe and welcoming at that kind of scale and diversity?
Parth Chadha: This is the hardest problem we work on, and the one I am most careful about. India is not one language with accents. Toxicity in Bhojpuri does not look like toxicity in Tamil, and a model trained mostly on English and Western data simply misses it, or worse, flags harmless regional speech as abuse. So the work is not just running moderation. It is making sure the moderation is fair across communities that have never been well represented in AI training data. We run this on Google’s Gemini models, fine-tuned heavily on Indian voice and Indian context, and that moderation work drove a 23% improvement in retention, because people stay where they feel safe.
Question: But are humans in the loop when it comes to AI voice moderation of voice rooms?
Parth Chadha: Two principles guide it. We keep humans in the loop on the hard calls rather than letting a model decide alone. And we treat a young user in a tier 3 town who is new to the internet as the person we protect first. Getting this wrong at our scale is not a bug. It is a harm, and we build like we believe that.
Question: A big part of your mission seems to be helping creators in places like Indore, Ranchi or Guwahati actually earn from their passion. How is AI helping those creators grow and monetise in ways that weren’t possible a few years ago?
Parth Chadha: A few years ago, if you were a talented gamer in Guwahati, the path to earning from it ran entirely through platforms that paid attention to metros and to English. You were invisible by default. Two things changed that. First, the platform itself, where a creator earns directly from their community. Second, AI lowered the skill barrier. A creator who never had an editor or a coach now gets AI feedback after every stream on what held the room and what did not. That used to be an advantage only well-resourced metro creators had.
Question: You’re backed by Google’s AI Fund and building on their models, at a moment when India is thinking hard about its own AI future. Where do you see the intersection of AI and gaming heading in India over the next few years, and how does STAN want to shape it?
Parth Chadha: For most competitive games today, AI is already coaching players, and not just on the mechanical side. It is starting to coach the mental side too. How you focus, how you recover from a bad game. For creators, AI shows up as a moderator or an editor. In our case, as a coach. The models under most of this for us are Google’s Gemini, tuned heavily for Indian languages. None of this is sudden. AI and gaming have been growing into each other for years, and that does two things at once: the skill ceiling rises for the best, and the learning curve gets gentler for everyone starting out.
As for shaping it, I would put it the other way around. STAN has never wanted to shape gaming. We want to be its home. And a home is not something you shape, it is somewhere you feel safe no matter how much you change. Gamers grow up. They switch titles, they go quiet for a year, they come back. We would rather be the place still standing, still in their language, when they do.
