Radar Makes Podcasts Searchable — and Usable by AI Agents
Particle, the AI newsreader startup from ex-Twitter engineers, is pivoting to index the spoken conversations buried in podcasts — and hedge funds are already paying for what their agents can't see.
Published: 2026-08-31 Category: Quick Take Sources: TechCrunch
The Pivot to Audio
Particle, the AI newsreader startup founded by former Twitter engineers, is shifting toward a more lucrative idea: indexing the spoken conversations buried in podcasts and making them discoverable. The company's new product, Radar, is a podcast search engine that doesn't just transcribe audio but understands its meaning, pulling out key quotes and highlights.
The business potential is already proving out. CEO and co-founder Sara Beykpour says hedge funds have been the highest-volume customers integrating directly with the API — drawn to data their AI agents can't otherwise see. Other top-paying customers include AI search platforms and data resellers, with Exa, the search-API provider for AI agents, among Radar's partners.
Why Audio Is a Blind Spot for Agents
The insight driving the pivot: most agents and services crawl the web and focus on text, but audio is a gap. "Agents are generally blind to audio; they can't see it unless something or someone has transcribed it," Beykpour said. Radar aims to provide that layer.
The product grew out of one of Particle's most-loved newsreader features — sourcing interesting podcast clips alongside related news. The team realized the value but also that it was trapped inside the news reader, so as the AI-agent wave accelerated, the company bet on an API-first podcast-intelligence product.
Scale and Features
Radar transcribes more than 130,000 podcasts — making it, per Particle, the largest transcribed podcast service in existence — including all the Apple Top 200 podcasts across 135 verticals, with 20,000 episodes added daily. Transcriptions carry speaker labels and rich metadata, understanding the entities (people, companies, brands, products, topics) being discussed.
It can track entity mentions across podcasts and send alerts via email, Slack or webhook, with filters to narrow search by guest or topic. It extracts self-contained, timestamped clips for both listening and reading, and tracks topics, listeners, ratings, reviews and ads. There's even a dedicated podcast ads search engine that finds every episode where a given company advertises and tracks trends over time. While all this is accessible via a web interface, the real product is the API and MCP server that lets AI agents tap in programmatically.
Analysis
Radar is a textbook bet on the "boring layer" of the AI-agent stack. The most valuable data isn't always the newest model — it's the structured access to content that was previously invisible. Podcasts are a vast corpus of opinion, expertise and corporate communication that text-focused crawlers simply can't reach, and being first to make them machine-readable is a defensible position.
The hedge-fund demand is telling. Funds that want an information edge over their own agents need proprietary, well-structured data — and transcribed, entity-tagged podcast intelligence is exactly that. It's also a reminder that the agent economy isn't only about consumer chatbots: it's about selling the raw ingredients (search, data, retrieval) that make agents useful in the first place. At $29 a seat, with custom API pricing and an ads-intelligence layer with clear monetization, Radar is a pragmatic, recurring-revenue take on the media-intelligence market — the kind of infra play that often outlives the flashier consumer bets that grab headlines.
Source: TechCrunch — "Radar makes podcasts searchable - and usable by AI agents" (Aug 26, 2026).