Particle, a startup established by ex-Twitter engineers, is channeling its efforts toward an innovative venture: cataloging spoken dialogues in podcasts to enhance their discoverability. The company recently unveiled Radar, a podcast search engine that not only transcribes audio but also comprehends its context, allowing for the extraction of significant quotes and highlights.
This initiative shows considerable commercial promise, already piquing the interest of hedge funds that seek unique data that is typically hidden from view, according to Particle’s co-founder and CEO, Sara Beykpour.
“Our most substantial clients have been hedge funds that are integrating directly with the API,” Beykpour mentioned. Journalists and researchers are expected to benefit from these tools, and other high-paying clients include AI search companies and data resellers, such as Exa, which collaborates with Radar.
The concept originated from one of the features of Particle’s news-reading application, which utilized an API to present intriguing podcast snippets alongside relevant news articles.
Recognizing the potential of this product, the team realized it was constrained within the news sector. As interest in AI agents grew, Particle opted to pivot and develop an API focused on podcast intelligence.
“Our aspiration is to encapsulate all new media intelligence and audio intelligence within this API. Most services and agents mainly focus on text while we aim to provide a comprehensive layer for audio,” Beykpour explained. “Many agents overlook audio unless it’s been transcribed.”
With Radar, the platform transcribes over 130,000 podcasts, claiming to be the largest transcribed podcast service available. It includes content from the Apple Top 200 podcasts across various categories, adding around 20,000 episodes to its index each day.
The transcriptions feature speaker identification and detailed metadata, as Radar comprehends the entities discussed—such as individuals, companies, brands, products, and topics.
Additionally, it can monitor mentions of these entities within podcasts and provide notifications whenever they are referenced, whether in real time or as daily or weekly summaries.
The notifications, which can be sent via email, Slack, or webhook, are customizable with filters. Users can configure Radar to only notify them when specific guests discuss certain topics or limit alerts to top podcasts only.
Radar can also extract relevant self-contained clips with timestamps, allowing users to both listen to and read the remarks made.
“We’ve curated notable clips, providing the perfect solution for those who cannot listen to the entire podcast or prefer to avoid summaries,” Beykpour added.
Radar tracks topics mentioned in podcasts, references to individuals and items, listener ratings, reviews, and even ad placements within episodes. Moreover, it features a dedicated search engine for podcast advertisements, tracking mentions of companies over time and offering valuable trends. This capability has additional monetization potential, including political bias assessments, ranking data, audience estimations, sponsorship insights, and brand suitability analyses.
While Radar offers a web interface, its primary product is the API and MCP, enabling businesses and AI agents access to this intelligence programmatically.
The pricing for Radar starts at $29 per month for individual users, with a business plan priced at $399 per month for 20 users. API users have a custom pricing structure based on their requirements.
Looking ahead, Radar plans to broaden its offerings beyond podcasts to include various audio sources, such as YouTube videos and news broadcasts.
