India-based audio series platform Pocket FM aims to become the Netflix of audio by matching its extensive library of episodes to users’ preferences. To achieve this, the company is leveraging AI to accelerate content production.
The startup, backed by Lightspeed, is equipping its writers with AI tools designed to enhance storytelling. These tools can suggest better episode endings, improve narrative engagement, and streamline the writing process. Pocket FM already uses AI solutions like ElevenLabs to generate voices for its audio series and has tested AI-assisted writing tools internally.
Founder Rohan Nayak explained that the AI tools, now available to all writers, will help them complete episodes faster. One such tool, called CoPilot, assists writers in transforming narrative-based content into dialogue-driven segments. It also performs “beat analysis” to tailor writing for specific genres, ensuring higher engagement. CoPilot includes standard writing features like shortening or expanding text and generating content from prompts.
To develop CoPilot, Pocket FM analyzed thousands of hours of user data to identify what makes certain storylines more compelling. The tool now offers suggestions to heighten character conflicts, recommend gripping endings, and propose background effects for audio production. Additionally, it generates character bios, tracks relationships, and summarizes plot points for easy reference.
CoPilot also includes a review function that checks grammar, plot consistency, and provides qualitative feedback. Behind the scenes, Pocket FM trains smaller AI models to maintain story context, character arcs, and narrative coherence. User feedback further refines the AI’s ability to add drama and tension to stories.
Alongside these AI advancements, Pocket FM has introduced localization tools to adapt content for different markets. These tools not only translate text but also adjust names and phrases to align with regional cultural preferences. The company first tested this feature in Germany, where it struggled to engage users last year. The trial proved successful, boosting monthly in-app revenue to over $700,000 by June.
Nayak noted that expanding into new markets previously took 12-18 months, but AI has reduced this timeline to under three months. In Germany, writer productivity increased by 50%, and error-free drafts led to higher user retention. In the U.S., AI-assisted shows now account for 10% of playtime and have generated $7 million in revenue over the past year while cutting production costs by two to three times.
Pocket FM’s AI adoption has enabled rapid content scaling, with nearly 1,000 pilots launched monthly. While most shows don’t succeed, the sheer volume ensures some become hits. Beyond audio, the company is exploring other formats, including comic strips through Pocket Toons and potential video adaptations. It has also raised over $196 million in funding and is experimenting with a micro-drama app.
By next year, Pocket FM plans to launch its own large language model (LLM) trained on its show data. This model will integrate writing assistance, adaptation, dramatization, and story continuity, eliminating the need for multiple smaller AI models.
However, AI adoption has brought challenges. Pocket FM has laid off employees and contractors in multiple rounds over the past year, with reports of diminishing returns for writers. The company also faces lawsuits in California over employment and wage disputes. A spokesperson stated that AI has minimally impacted core creative teams, instead expanding opportunities for output and reach.
Quality concerns persist, as AI-generated content risks flooding the platform with low-quality material. Pocket FM counters this by using AI moderation to check for duplication, copyright issues, and content health before approval. Shows are ranked based on user engagement, ensuring popular content rises to the top.
Another issue is writers’ growing reliance on AI. In Germany, AI-generated content now surpasses human-written material for some titles. As AI tools proliferate, expectations for higher output may rise, potentially lowering average returns unless user growth keeps pace.
Pocket FM did not directly address concerns about declining returns but emphasized that AI helps writers refine episodes efficiently. By focusing on targeted edits rather than full rewrites, writers can maintain quality while increasing productivity. The company believes well-structured stories will thrive, regardless of AI assistance.