Particle, the AI newsreader startup founded by former Twitter engineers, has shifted its product focus toward indexing and making searchable the spoken content in podcasts. On Wednesday the company announced Radar, a podcast search engine that not only generates transcripts but also interprets spoken content to surface key quotes, topics and highlights.
Why this matters
Podcasts contain large volumes of spoken information that standard text-focused crawlers and agents typically cannot access. According to Particle co-founder and CEO Sara Beykpour, Radar has already attracted commercial interest from hedge funds seeking data that their human agents miss. Beykpour told TechCrunch that hedge funds are among the highest-volume customers directly integrating with the API. Other paying customers include journalists, researchers, AI search platforms and data resellers; Exa, a search API provider for AI agents, is cited as one of Radar’s partners.
What Radar does
Radar provides more than simple speech-to-text: Particle says it transcribes over 130,000 podcasts, making it the largest transcribed podcast service the company is aware of. The index covers all Apple Top 200 podcasts across 135 verticals, and roughly 20,000 episodes are added to Radar’s index every day.
Transcripts include speaker labels and rich metadata. Radar performs entity recognition for people, companies, brands, products and topics, and it can track mentions of those entities across podcasts and flag when they occur.
Alerts, search filters and clip extraction
Users can receive alerts when specified entities are mentioned, either immediately or as daily or weekly digests. Alerts can be delivered via email, Slack or webhook, and are configurable with filters — for example, to only notify when a particular guest appears and discusses a specific topic. Searches can also be limited to subsets such as top podcasts.
Radar can extract self-contained clips with timestamps so users can both listen to and read the exact segments. Particle offers pre-selected notable clips for users who prefer quick audio excerpts to full-episode listening or longer summaries.
Additional intelligence and monetization features
Beyond transcripts and clips, Radar tracks topics, mentions, listener ratings and reviews, episode ads and other metadata. There is a dedicated podcast ads search engine that finds each episode where a specific company advertises and tracks trends over time. Additional product features with monetization potential include political bias analysis, chart ranking data, audience size estimates, sponsorship data and brand suitability assessments.
Product model and pricing
While Radar is accessible through a web interface, Particle emphasizes that the core product is the API and MCP, which allow AI agents and businesses to programmatically tap the same intelligence. Radar’s consumer-facing pricing starts at $29 per month per seat; there is a $399-per-month business plan that includes 20 seats. API customers receive custom pricing based on their usage requirements.
Roadmap
Particle says it plans to extend Radar beyond podcasts to other audio sources such as YouTube videos and news clips in future updates.
In short, Radar aims to expose audio-based media intelligence through an API, filling a gap where most agents and services remain effectively blind to raw audio unless it has been transcribed and semantically indexed.
