India based in the field of the Audio series platform FM pocket money It is to be Netflix Audio. This means that the company intends to match its audio series to hundreds of episodes to the tastes of users. For this to work, he must quickly free the content – something that he now turns to AI.
The startup supported by the light gives writers a set of AI tools that can do things such as suggesting better ends of the episode or make a more addictive narrative. We hope that the tools will speed up the history of writing.
Pocket FM already uses some AI tools, such as ElevenLabs to generate votes for the audio series. He also tested AI tools for writing and internal adaptive assistance.
Rohan Nayak, the founder of Pocket FM, said he was introducing AI tools for all writers, so the end of the episodes would take them less time.
Writing tool, named CopilotIt can be used to help every writer create a story.
Copilot can transform the narrative based on writing based on dialogues for a specific segment. He can also perform a “beats analysis” to shape the letter in such a way that it is more involved in the audio series of a specific species. The tool also has basic chatbot -style writing functions, such as “Shorten”, “Development” and the ability to generate text using monitors.
To build Copilot, the company has examined thousands of hours of data points to understand what makes users more engage in a specific story in a specific genre.
On this basis, he added functions of writing suggestions designed to increase conflicts between characters and recommend endings for the episode to be more exciting. AI may also suggest background effects that can be used during sound production.
The tool can automatically generate the bios of characters, their relationships and summarize the story points of different episodes, enabling creators to return to these details while writing.
Copilot also has a review tool that checks feature points, grammar and leaves qualitative feedback through comments about the episode.
Under the hood of Pocket FM, he trains smaller models to maintain the context for the history of arches and character relationships, as well as narrative consistency. In addition, using signals from users, the startup destroys artificial intelligence to add more drama to history.
International expansion and location plans
In addition to the arrival of AI tools, Pocket FM has launched adaptation tools to various markets, which not only explain the text from one language to another, but also change names and phrases that are more suitable for the culture of this region.
The company debuted for the first time with a tool as part of the Copilot package in Germany at the beginning of this year to convert stories from other regions after apparently Fighting for users' involvement in a European country last year.
Nayak said that the company recorded great results from this process, with a steady increase in monthly revenues in the application, which in June exceeded USD 700,000.
“When we began to develop for new regions, it existed in this market for 12-18 months. You must have at least 1000 hours of content to start acquiring users and scaling the market. Now we can do it in less than three months,” said Nayak.
The tool has increased the writer's performance by up to 50% for the German market in terms of performance. In addition, the tool has helped the company create more flawless sketches of programs that caused a higher user retention for the audio series.
In the United States, the series created with the help of these new AI tools now contributes to 10% of the game time. In addition, these programs generated $ 7 million in revenues in the last 12 months, while reducing production costs by 2-3 times.
Building technology for scaling content generation
As a result of the internal adoption of various AI functions, Pocket FM was able to scale the content quickly. The startup said that it introduces nearly 1,000 pilots per month. And the very amount of content means that several of them become hits.
But the audio program is only one part. The company is already working on tools for conversion of history into comic stripes with Pocket money platform. In addition, Nayak said that a video is a possible format that the company can also explore. The startup, which collected over 196 million dollars of financing during rounds, also experiments with a micro dramatic application.
Next year, Pocket FM wants to publish its own large language model (LLM), which will be based on data collected from shows and contains various tools, such as writing help, adoption, dramatization and stopping of the context of history. Co -founder of the company, Pratek Dixit, said that when he passes to his own LLM, he would not have to train a lot of small models for separate functions.
Potential disadvantages of AI
AI's reception had its side effects.
Pocket FM has already dismissed people who were employees or contractors By many round In the last 12 months. There were also reports that the writers see reduced returns overtime. And the company is facing lawsuits in California regarding problems related to employment and payroll.
“Like most industries run by content, we work with a variety of network of writers, voice artists and production partners on a project basis, adapting resources to each market. AI had a minimal impact on our basic creative community; instead, it opened new possibilities of expanding coverage and production,” said the company's representative in response to these systems.
There are also quality questions. The company measures the quality according to the number of program storage.
The basic argument is that the new AI tools act as a room for writers even for Solo creators, so they will be able to produce more content at a fast pace. In addition, based on numbers, writers can quickly edit history with the help of artificial intelligence. However, these tools can very well induce “AI Slop”-a low-generated AI-to Platform content and can affect the user's recommendations, which makes it difficult to discover good stories.
Pocket FM claims that stories about a solid structure will gain popularity, even though AI has helped them.
The company has noticed that each element of content is checked by AI moderation frames to ensure quality and originality. He also claims that checking the moderation of AI such as duplication, copyright problems, content of content and other quality measures before approving the sound to start. Each program receives an equal push, and users' involvement ultimately determines the program ranking.
Another problem is that writers can become too dependent on artificial intelligence over time.
In Germany, AI writes more content than people for a program for selected titles. Thanks to Pocket FM plans to enter more AI, the amount of content written by AI may increase. And thanks to this, waiting for more programs can also increase. While the adoption of users also increases rapidly, the average phrases may fall.
The company did not directly concern TechCrunch's questions on phrases, but she said that her AI tools can accelerate the writer's work and help them edition of the episode based on the numbers and opinions of recipients. This means that they can introduce targeted improvements instead of full rewrite.
“In this way, faster creating of content does not necessarily not necessarily expand the quality or meaning; he simply changes the role of the writer in edition, refining and control of a more productive output,” said the spokesman in a statement.
We always try to evolve and provide insight into your perspective and opinions in TechCrunch and our coverage and events, you can help us! Bloat This survey To inform us how we are doing and have a chance to win the prize in return!
