The archive · AI & Models · Strategic decision · 2024–2026
Particle pivots AI newsreader into Radar: podcast search API for agents and hedge funds
Ex-Twitter-engineer startup Particle pivots to Radar: an API over 130,000+ transcribed podcasts for AI agents and hedge funds.
Particle · Radar
What the business is
Podcast intelligence platform: transcribes and indexes more than 130,000 podcasts with entity tracking, alerts, clip extraction, and an API/MCP that lets AI agents and businesses query spoken conversation.
How it started
Particle was founded by former Twitter engineers as an AI newsreader. Its app used an internal API to find interesting podcast clips and show them with related news stories; the team realized the technology was trapped inside the consumer app as the agent wave grew.
What happened
In August 2026 Particle launched Radar as its main product: transcribes 130,000+ podcasts (claiming the largest transcribed podcast service), including all Apple Top 200s across 135 verticals, ~20,000 episodes a day, with speaker labels, entity tracking, alerts via email/Slack/webhook, and a dedicated podcast-ad search engine. Hedge funds became the highest-volume API customers; Exa is a partner. Pricing is $29/month per seat and $399/month for 20 seats, with custom API pricing.
How it ended up
Live and scaling: the web interface, API and MCP are shipping; Particle plans to extend Radar beyond podcasts to YouTube videos and news clips.
Background
Particle is an AI newsreader startup founded by former Twitter engineers. Buried inside the app was an internal API that found notable podcast clips and surfaced them alongside related news stories — a feature users loved but that was locked inside the consumer product. As the AI-agent wave grew, the team saw a bigger market: agents crawl text and are blind to audio unless someone transcribes it first.
In August 2026 Particle pivoted and launched Radar as its core product: a podcast search engine that transcribes and analyzes more than 130,000 podcasts — which it says is the largest transcribed podcast service in existence — including all Apple Top 200 podcasts across 135 verticals, with around 20,000 episodes added daily. Transcripts carry speaker labels and entity metadata, and Radar tracks people, companies, brands and topics, sending alerts by email, Slack or webhook.
The real product is the API and MCP rather than the web interface: hedge funds have been the highest-volume customers integrating Radar directly, with AI-search platform Exa among the partners. Radar is priced at $29 per month per seat, $399 per month for a 20-seat business plan, and custom pricing for API users. Particle plans to expand beyond podcasts into other audio, including YouTube videos and news clips.
What has to be true
- Audio was an unclaimed layer: every agent and crawler indexes text, so transcribed, searchable podcasts were a gap worth owning before a platform did.
- The pivot repackaged proven demand: the podcast-clip feature was already popular inside the newsreader, so Radar launched with validated product-market pull rather than a guess.
- Enterprise buyers made the economics real: hedge funds integrating via API were willing to pay for proprietary data no text crawler can reach.
- The API/MCP positioning turns a search engine into infrastructure: distribution happens through agents and businesses, not a consumer app competing for attention.
What can be applied
When a valuable capability is trapped in a consumer product, spin it out as an API: the same intelligence that made a feature beloved becomes a standalone product once it has its own distribution.
Aftermath
As of late August 2026 Radar is Particle's main product after the pivot away from the consumer newsreader focus. It indexes 130,000+ podcasts with ~20,000 episodes added daily, and its highest-volume paying customers are hedge funds using the API; AI-search provider Exa is publicly named as a partner. Pricing runs $29/month per seat and $399/month for 20 seats, with custom API terms. Particle says the next step is expanding Radar to other audio forms such as YouTube videos and news clips, keeping the same intelligence layer for agents.
Sources
- Radar makes podcasts searchable — and usable by AI agents
- This AI Search Engine Can Find Anything Said in Podcasts
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card