Particle, an AI-driven news reader company founded by ex-Twitter engineers, is redirecting its efforts towards a potentially more profitable venture: cataloging the spoken discussions embedded within podcasts and making them searchable. This Wednesday, the firm unveiled Radar, a podcast search engine capable of not only transcribing audio but also comprehending its meaning, thereby enabling the extraction of significant quotes and key moments.
This innovation holds considerable business promise, having already garnered attention from hedge funds seeking unobservable data for their agents, according to Particle co-founder and CEO Sara Beykpour.
Beykpour informed TechCrunch that “Hedge funds represent the primary high-volume clients directly integrating with our API.” Although journalists and researchers might also find these tools useful, other significant revenue generators include AI search platforms and data reselling services. For example, Exa, an AI agent search API provider, is one of Radar’s collaborative partners.
The concept itself originated from a highly popular feature within Particle’s news-reading application. This app leveraged its API to identify compelling podcast segments, which it subsequently integrated into the app’s feed alongside relevant news articles.
The team at Particle recognized the inherent worth of this product, yet also noted its limited scope within the news reader. With the increasing momentum surrounding AI agents, the company chose to pivot, concentrating on developing an API for its podcast intelligence solution.
Beykpour stated, “Our ambition is truly to consolidate all new media intelligence and audio intelligence within this API. This domain is particularly compelling because most API agents and services primarily scour the web, concentrating on text. We are introducing this audio layer.” She added, “Agents typically cannot process audio; it remains inaccessible to them unless it has been transcribed.”
Through Radar, the company transcribes over 130,000 podcasts, establishing itself as the most extensive transcribed podcast service available. This encompasses all Apple Top 200 podcasts across 135 different categories, with an additional 20,000 episodes indexed by Radar each day.
The podcast transcriptions feature speaker identification and comprehensive metadata, since Radar comprehends the entities — such as individuals, businesses, brands, products, and subjects — that are mentioned.
Furthermore, it can monitor instances where these entities are mentioned across various podcasts, dispatching alerts either in real-time or as daily or weekly summaries.
These alerts, deliverable via email, Slack, or webhook, offer customizable filtering options. Users can, for example, configure Radar to issue alerts exclusively when specific guests appear and address a particular subject. The search functionality also allows for further refinement, such as restricting results to only top-tier podcasts.
Radar possesses the capability to extract pertinent, self-contained audio clips, complete with timestamps, enabling users to both hear and read the spoken comments.
Beykpour explained, “We’ve curated significant clips; thus, if you lack the time to listen to an entire podcast or prefer not to read a summary, this offers an optimal method to grasp the podcast’s content.”
Additionally, Radar can monitor the subjects discussed within a podcast, identify who or what was referenced and at what point, track listener ratings and reviews, analyze episode advertisements, and more. A specialized podcast ad search engine is also available, which can pinpoint every episode where a specific company advertises and observe its historical trends. This capability, along with other features like political bias assessment, chart ranking data, audience size projections, sponsorship insights, and brand suitability evaluations, presents further opportunities for monetization.
Although all these functionalities are accessible via Radar’s web interface, its core offering resides in its API and MCP, enabling AI agents and other enterprises to programmatically access this identical intelligence.
Radar is available for $29 monthly per user, with a business plan priced at $399 per month covering 20 users. API access for users is subject to customized pricing, tailored to their specific requirements.
Looking ahead, Radar intends to extend its service beyond podcasts to incorporate other audio formats, including YouTube videos and news segments.