Radar makes podcasts searchable — and usable by AI agents
Particle, the AI newsreader startup founded by former Twitter engineers, has launched Radar, a search engine that transcribes and semantically indexes podcast conversations, making spoken audio content discoverable and usable by AI agents. Chief executive Sara Beykpour said the pivot addresses a gap in the market, since most AI crawlers and agents focus on text and remain effectively "blind" to audio unless it has been transcribed. The tool has already drawn strong interest from hedge funds seeking data that automated systems cannot otherwise access, alongside journalists, researchers and AI search platforms.
Radar transcribes more than 130,000 podcasts, including the entire Apple Top 200 across 135 categories, with 20,000 new episodes added daily, making it the largest transcribed podcast service available. Transcripts include speaker labels and detailed metadata identifying people, companies, brands and topics, allowing Radar to track mentions across shows and send customisable alerts via email, Slack or webhook. The platform also surfaces notable timestamped clips, tracks listener ratings and reviews, and includes a dedicated search engine for podcast advertisements.
- Particle launches Radar, an AI-powered podcast search and transcription engine
- Hedge funds are its biggest customers, seeking data agents can't otherwise see
- Indexes 130,000+ podcasts with entity tracking, alerts and clip extraction
New here? Start with this
Podcasts are audio programmes made up of episodes, often featuring interviews or discussions. Because their information is spoken rather than written, it can be harder for search tools and automated systems to find particular comments or follow a topic across different shows.
Particle is a news technology company founded by former Twitter engineers, and Sara Beykpour is its chief executive. Hedge funds, journalists and researchers are among the groups interested in finding information in podcasts, while AI agents are software systems that can search for and use information on a person’s behalf.
Turning speech into text can make podcast conversations easier to search and organise. It also raises questions about how spoken material, including comments and advertisements, can be tracked and used by people and automated systems.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Radar solves a genuine problem by making publicly distributed podcast content discoverable and accessible for research, journalism, and information discovery. Podcasters intentionally published their work for broad audiences, and transcribing and indexing this public content extends its reach without preventing creators from monetising through advertising, sponsorships, or premium subscriptions. The tool democratises access to information and enables better research capabilities across thousands of public conversations that were previously scattered and difficult to find.
The case against
Whilst podcasts are publicly distributed, creators published them as audio without necessarily consenting to automated transcription, semantic indexing, and unrestricted searching by commercial systems and hedge funds seeking financial intelligence. Creators lose control over how their intellectual property is indexed and used, with no mechanism to benefit from or consent to the value extraction that occurs when their conversations become machine-searchable data. The tool raises questions about fair compensation and whether public availability should mean unlimited algorithmic access and analysis, particularly when used for commercial gain by entities other than the creators.