Pocket FM

Big milestone for Pocket FM on crossing $400M ARR. What excites me most is not just the number, but how it’s being built. Audio is no longer passive consumption, it’s a habit. Users are spending 140+ minutes a day, coming back with real intent. Monetization is not reliant on fragile autopay. It’s driven by engagement and microtransactions, making the model far more resilient. This category is no longer niche. With 100B+ minutes consumed annually, audio is becoming a default daily behavior. And this isn’t a single-market story. It’s a truly global play, with markets like Germany scaling rapidly alongside the US and India. Feels like we’re still very early. 🚀

Pocket FM has now crossed $400M ARR. The first $200M took 6 years, the next came in just 12 months. AI changed everything for us. When we started Pocket FM , the idea was to build a new entertainment format - bingeable, bite-sized audio series. We built this category in India, then scaled it to the US and Europe and today it’s part of mainstream entertainment across 20+ countries. While building Pocket, we reimagined the entertainment playbook by pivoting to an AI-native storytelling system. At the core of this AI-native engine is our fiction writing co-pilot, trained on billions of minutes of engagement data. It enables creators to go from a raw idea to a fully dramatized series in minutes, while allowing us to rapidly identify potential blockbusters and adapt stories across languages and cultures almost instantly. Content creation has exploded on Pocket - 300,000+ creators are now producing 80,000+ hours of content every month and this isn’t low-effort content; it’s storytelling people keep coming back to every day. AI isn’t replacing creativity, it’s unlocking it. One creator story recently stuck with me, a first-time creator from Hyderabad whose show found an audience in the US and he made ~$50,000 (₹50 lakh) in a single month. This is democratization of storytelling. Feels like this is just Episode 1. Many more cliffhangers ahead. P.S. We are now free cash flow positive, at ~5% EBITDA.

A single creator can now pull off the kind of storytelling that traditionally needed writers' rooms, studios, and months of production. And we have 300K+ creators doing exactly this at Pocket FM . When AI scales and assists the craft, and creators lead the vision, the output is staggering. A jump from $200M to $400M+ ARR in 12 months says it all. 🚀

Oh wow…Pocket FM’s AI magic catapults from $200M to $400M ARR in just 12 months, unlocking creator goldmines worldwide! :)

Pocket FM 's AI Product team is building narrative AI infrastructure at scale + actively looking for people at the intersection of computational narratology, linguistics and applied AI. You might be a fit if your background is in: - narrative theory, screenwriting, or creative writing - computational narratology, or NLP/ML with a strong grounding in storytelling - computational linguistics, knowledge representation, or formal semantics - a combination of the above https://lnkd.in/dyb5fnMa

We're hiring! Pocket FM 's AI Product team is looking for storytellers with LLM expertise as well as interns for our operations team. Apply here: https://lnkd.in/deCyr2yN

𝗠𝗼𝘀𝘁 𝗔𝗜 𝗶𝗻 𝗰𝗼𝗻𝘀𝘂𝗺𝗲𝗿 𝗮𝗽𝗽𝘀 𝗶𝘀 𝘀𝘁𝗶𝗹𝗹 𝗯𝗹𝗶𝗻𝗱 𝘁𝗼 𝗰𝗼𝗻𝘁𝗲𝘅𝘁. We saw this firsthand at Pocket FM . Last week, I mentioned we will start sharing more about the AI systems we are building. Here is one for this week. We replaced our comment analysis system. Old system: approximately 60 percent accuracy New system: more than 87 percent accuracy Sounds like a routine upgrade. It was not. Our earlier models were trained on social media data. So when a user said: “I hate this villain” The system marked it as negative. But in storytelling, that is one of the highest compliments you can get. The model was not wrong. It just did not understand the world it was operating in. 𝗦𝗼 𝘄𝗲 𝗿𝗲𝗯𝘂𝗶𝗹𝘁 𝗶𝘁. We created a Comment Sentiment Agent that understands stories: It sees the character canvas of every show It processes episodic summaries as context It interprets sentiment through narrative, not just language But the real unlock came after. We categorized every comment into what actually matters — Plot, Characters, Production, Platform. Then we connected sentiment directly to episode retention. Now we do not ask: Why are users dropping off at episode 170? We can say: The protagonist's arc stalled after episode 167. That is where the drop-off starts. Fix that. That is the shift AI needs. 𝗡𝗼𝘁 𝗺𝗼𝗿𝗲 𝗴𝗲𝗻𝗲𝗿𝗶𝗰 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲. 𝗠𝗼𝗿𝗲 𝗰𝗼𝗻𝘁𝗲𝘅𝘁𝘂𝗮𝗹 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲. We are building systems like this every week. Not to showcase AI. But to solve very specific problems, deeply. Because the best AI is not the flashiest. It is the one that actually tells you what to fix. Prateek Dixit Lalit Gangwar Ashu Behl Nitin Verma Vineet Singh Nishant KS Akansha Kumari