Particle is moving from AI news reading toward a more targeted problem: making podcast conversations searchable, trackable, and usable by software systems. Its new product, Radar, turns spoken audio into structured podcast intelligence for people and businesses that need to find what was said, who said it, and where it happened.
The company is presenting Radar not only as a web search tool, but as infrastructure for AI agents. That matters because many agents and automated research systems are built around text from the open web, while audio remains harder for them to access unless it has been transcribed and organized.
What Radar Does
Radar is a podcast search engine from Particle, the AI newsreader startup founded by former Twitter engineers. The product transcribes podcast audio, identifies meaning in the conversations, and pulls out important quotes and highlights.
According to the source article, Radar transcribes more than 130,000 podcasts. That includes all the Apple Top 200 podcasts across its 135 verticals, with 20,000 episodes added to Radar's index daily.
The value is not limited to raw transcription. Radar adds speaker labels and metadata, and it recognizes entities being discussed, including people, companies, brands, products, and topics. That makes the system useful for more than keyword matching, because users can track what is being discussed across many podcast episodes.
For a user trying to follow a company, person, product, or topic, that changes the role of podcasts from long-form media into searchable source material. Instead of listening through full episodes, a user can find relevant moments, review clips, and monitor future mentions.
Why AI Agents Need Audio
Particle co-founder and CEO Sara Beykpour told TechCrunch that hedge funds have shown strong demand for Radar because they want data their agents cannot already see. She said hedge funds have been the highest-volume customers directly integrating with the API.
Other paying users have included AI search platforms and data resellers. Exa, described in the source article as a search API provider for AI agents, is among Radar's partners.
The business logic is straightforward. If AI agents rely heavily on text, then podcast audio is an information source that can be missed. Radar is designed to provide an audio layer that can be queried, filtered, and connected to other systems through an API and MCP.
That distinction is important for the product. Radar's web interface is available, but the source article says the real product is the API and MCP. Those interfaces let AI agents and other businesses access the same podcast intelligence programmatically.
From News Feature To Standalone Product
The idea behind Radar came from Particle's own news-reading app. The app had used an API to source interesting podcast clips and place them next to related news stories in the feed.
Particle's team saw that this feature had value outside the news app. As interest in AI agents grew, the company decided to shift its focus and build an API around podcast intelligence.
This move turns a feature into a standalone product. In the newsreader, podcast clips supported the user experience around stories. In Radar, the podcast index itself becomes the main asset, with search, alerts, clips, metadata, and integrations built around it.
That pivot also changes the customer base. Instead of serving only readers inside one app, Radar can be used by researchers, journalists, AI search platforms, data resellers, hedge funds, and other businesses that want structured access to podcast conversations.
Alerts, Clips, Ads, And Metadata
Radar can track mentions of entities across podcasts and send alerts when they appear. Alerts can arrive when a mention occurs, or they can be delivered as a daily or weekly digest.
The alerts can be sent through email, Slack, or webhook. Users can also apply filters, such as limiting alerts to cases where certain guests appear and discuss a particular topic. Searches can also be narrowed by podcast category, including limiting results to top podcasts only.
Radar also extracts self-contained clips with timestamps. That allows users to read or listen to the relevant section without working through an entire episode.
The product tracks several other podcast signals as well, including topics mentioned, who or what was mentioned and when, listener ratings and reviews, episode ads, and more. One dedicated feature is a podcast ads search engine that can find every episode where a given company advertises and track how that changes over time.
The source article also describes additional monetization potential around political bias analysis, chart rankings data, audience size estimates, sponsorship data, and brand suitability.
Pricing And What Comes Next
Radar is priced at $29 a month per seat. A business plan costs $399-per-month and includes 20 seats. API users receive custom pricing based on their needs.
For now, Radar is focused on podcasts. In the future, Particle plans to expand the service beyond podcasts to support other forms of audio, including YouTube videos and news clips.
That direction fits the larger idea behind the product: audio contains useful information, but it needs to be indexed before people, companies, and AI agents can work with it efficiently. Radar is Particle's attempt to turn podcast conversations into searchable, alertable, and machine-usable data.