Pocket FM Deploys AI Tool to Assist Writers with Story Development and Cliffhangers

Indian audio series platform Pocket FM aspires to become the Netflix of audio entertainment. Its strategy involves matching its extensive library of multi-episode audio series to user preferences. To achieve this, the company must produce content at a rapid pace — a challenge it is now addressing with artificial intelligence.

The Lightspeed Ventures-backed startup is equipping its writers with an AI toolkit designed to enhance narratives, such as by proposing more compelling episode endings or increasing overall engagement. The goal is to significantly accelerate the storywriting process.

Pocket FM already leverages AI tools like ElevenLabs for voice generation in its audio series. It has also conducted internal trials with AI for writing and adaptation support.

Founder Rohan Nayak stated that by rolling out these AI tools to all writers, the company aims to reduce the time required to complete episodes.

The writing tool, named CoPilot, assists writers at any skill level in crafting stories.

CoPilot can convert narrative prose into dialogue for specific scenes. It also performs "beat analysis" to tailor the writing for greater engagement within a particular audio series genre. Additionally, the tool includes standard chatbot-style functions like "shorten," "expand," and text generation from prompts.

To develop CoPilot, the company analyzed thousands of hours of data to identify what drives user engagement with storylines across different genres.

Based on these insights, it incorporated features that suggest ways to heighten character conflict and recommend episode endings for added excitement. The AI can also propose tags for background sound effects to be used during audio production.

The tool automatically generates character biographies, outlines their relationships, and summarizes plot points from various episodes, enabling creators to easily reference these details during writing.

CoPilot includes a review function that checks for plot consistency and grammar, and provides qualitative feedback through comments on each episode.

Technically, Pocket FM is training smaller AI models to maintain story context, including character arcs and narrative consistency. Furthermore, by utilizing user engagement signals, the startup guides the AI to inject more drama into stories.

International expansion and localization plans

Alongside its AI toolkit, Pocket FM has launched adaptation tools for international markets. These tools not only translate text but also localize names and phrases to better align with regional cultures.

The company first introduced this tool as part of the CoPilot suite in Germany earlier this year, following challenges in engaging users there last year. It is used to convert stories from other regions for the local audience.

Nayak reported strong results from this pilot, with monthly in-app revenue showing consistent growth and surpassing $700,000 in June.

"When we began expanding into new regions, it typically took 12 to 18 months to establish a meaningful presence. You need at least 1,000 hours of content to start acquiring users and scaling the market. Now we can achieve this in under three months," Nayak explained.

For the German market, the tool boosted writer productivity by up to 50% in terms of show output. It also helped produce more error-free drafts, leading to higher user retention for audio series.

In the U.S., series created with these new AI tools now account for 10% of total playtime. These shows have generated $7 million in revenue over the past 12 months while reducing production costs by two to three times.

Building tech to scale content generation

By integrating various AI features, Pocket FM has accelerated its content production. The startup now releases nearly 1,000 pilot episodes monthly. This high volume increases the likelihood of producing hits.

However, audio shows are just one focus. The company is already developing tools to transform stories into comic strips via its Pocket Toons platform. Nayak also mentioned video as a potential future format. The startup, which has raised over $196 million, is additionally experimenting with a micro-drama app.

By next year, Pocket FM plans to launch its own proprietary large language model (LLM). This model will be trained on data from its shows and will integrate various functions like writing assistance, adaptation, dramatization, and story context management. The company stated that using its own LLM will eliminate the need to train numerous smaller models for individual features.

AI's potential downsides

Adopting AI has not been without consequences.

Pocket FM has conducted several rounds of layoffs affecting employees and contractors over the past year. Reports also indicate some writers have seen diminishing returns over time. The company is facing lawsuits in California related to employment and wage issues.

"Like most content-driven industries, we collaborate with a diverse network of writers, voice artists, and production partners on a project basis, adjusting resources per market. AI has had minimal impact on our core creative community; instead, it has created new opportunities to expand our reach and output," a company representative stated regarding the layoffs.

Quality is another concern. The company measures quality primarily through a show's user retention numbers.

The core idea is that these AI tools act as a virtual writers' room, enabling even solo creators to produce more content quickly. Writers can also use AI to make swift edits based on performance data. However, there is a risk of "AI slop" — low-quality, AI-generated content — flooding the platform, which could affect user recommendations and make it harder to discover well-crafted stories.

Pocket FM contends that stories with a strong foundational structure will succeed in popularity, regardless of AI assistance.

The company emphasized that all content is reviewed by its AI-powered moderation system to ensure quality and originality. This system checks for duplication, copyright issues, content health, and other quality metrics before an audio series goes live. Every show receives equal promotional support initially, with user engagement ultimately determining its ranking.

Another worry is writers becoming overly reliant on AI.

In Germany, for certain titles, AI is now writing more content per show than human writers. As Pocket FM introduces more AI tools, the proportion of AI-generated content may increase. This could also raise expectations for faster show production. Unless user growth keeps pace, average returns for writers could decline.

The company did not directly address questions about writer returns but stated that its AI tools accelerate a writer's workflow and help them refine episodes based on data and audience feedback. This allows for targeted improvements rather than complete rewrites.

"In this way, faster content creation doesn't inherently reduce quality or relevance; it shifts the writer's role more towards editing, refining, and guiding a more productive output," a spokesperson explained.

We are constantly striving to improve. By sharing your perspective and feedback on TechCrunch's coverage and events, you can help us evolve. Complete this survey to let us know how we're doing and enter for a chance to win a prize!