AI Particle is shifting the focus of its operations to audio data. The company has launched a podcast search product called Radar, with the objective of converting the language content previously dispersed among podcasts into searchable, accessible data and making it more accessible to AI agents and business clients.
Turned to podcast intelligence API
Particle was created by former Twitter engineer. According to the company, Radar does not simply translate podcasts into words, but also identifies the individuals, companies, brands, products and topics it mentions, and extracts key quotations and highlights.
This direction is derived from a function of Particle's original news applications. Previously, the application embedded relevant podcasts in the news stream. The team then determined that this part of the capacity would be more commercially available as a separate product, especially after AI agents began to need more non-textual data.
Hedge funds and AI platforms are connected
Particle Co-founder and CEO Sara Beykpour indicated that the one type of customer with the highest level of access is a hedge fund. Such clients would like to have audio information that their AI agents would not normally cover.
In addition to hedge funds, paid clients include AI search platforms and data resellers. TechCrunch mentions that Exa, which provides a search for API to AI agents, is already one of Radar's collaborators.
Covered 130,000 podcasts
According to the company, Radar has recorded more than 130,000 podcasts and described them as currently the largest podcast transfer service. Its index covers the top 200 out of 135 short-listed programmes with about 20,000 new editions per day.
Radar provides voice-labeled transcription text and metadata. It also tracks references by entity and sends a reminder when the content appears. Reminders include mail, Slack and webbook, and users can also set filter conditions, such as tracking only the content of a particular subject for a specific guest.
You can extract clips and track ads.
In addition to full-text retrieval, Radar can also extract independent clips with timetamps to facilitate direct listenership or reading of focus content. According to the company, such pre-screened footage is suitable for users who do not want to complete a podcast.
The product also tracks the placement of advertisements in the programme, the ratings and reviews of the audience, the ranking of the lists, the estimation of the size of the audience and the data on sponsorship. Particle also has a separate podcast ad search function, which is used to find out in which programmes a company advertises and to observe trends.
Core products are API and MCP
While Radar provides a web interface, Particle values API and MCP interfaces to allow AI agents and enterprise systems to access these audio data capabilities in a programmed manner. The company is currently charging $29 per month for individual users and $399 per month for the enterprise version, including 20 seats; the API clients are priced on demand.
Particle also plans to expand the reach of Radar from podcasts to more audio content, including YouTube videos and news audio clips.