Radar: A Search Engine That Turns Podcasts Into AI-Readable Data

Information-rich conversations within podcast episodes have long remained locked inside audio files, out of reach for search engines and AI agents that mostly rely on written text. But a startup founded by former engineers at «Twitter» has decided to address this gap through a new tool called «Radar», which promises to convert spoken words in podcasts into structured, searchable, and analyzable data.

From News Reader to Audio Search Engine

«Particle» began its journey by developing a smart news-reading app, but it recently chose to pivot toward an idea that seems more profitable: indexing podcast content and making it discoverable. The «Radar» project stems from a feature that was beloved in its original app, which displayed interesting audio clips alongside related news stories.

The team recognized the value embedded in this feature but noticed it remained confined within a news-reading app. With the rise of the AI agents wave, the company decided to redirect its compass toward building a specialized API (Application Programming Interface) for extracting information from audio.

CEO and co-founder Sara Beykpour explains that the company's vision revolves around gathering all «audio media intelligence» into a single API, noting that most current agents and services crawl the internet and focus solely on text, while remaining «blind» to audio content unless someone transcribes it into written text.

How Does Radar Work?

«Radar» doesn't just transcribe audio into text; it seeks to understand its meaning, enabling it to extract important quotes and key moments. According to the company, more than 130,000 podcast programs have been transcribed, making it the largest service of its kind, with about 20,000 episodes added to its database daily, including all of the top 200 programs in the «Apple» store across 135 categories.

The transcription processes include speaker identification and rich metadata, as the system recognizes mentioned entities such as people, companies, brands, products, and topics. It can also track mentions of these entities across various programs and send instant alerts or daily and weekly summaries.

Among the most notable features «Radar» offers:

• Customizable alerts via email, «Slack», or Webhook.
• Precise filters that allow receiving notifications when a specific guest discusses a particular topic.
• Extracting standalone clips with timestamps to listen to or read directly.
• A dedicated search engine for podcast ads that captures every episode in which a company advertises and tracks its performance over time.

The tool adds additional analytics covering political bias, chart classifications, audience size estimates, sponsorship data, and how suitable the content is for a brand.

A Promising Market and Unexpected Customers

Beykpour reveals that hedge funds were among the largest customers to integrate directly with the API, seeking data their agents couldn't access through traditional means. Customers also included AI-powered search platforms and data distributors, in addition to journalists and researchers.

Although the service is available through a web interface, the real product lies in the API and the MCP protocol, which allow AI agents and companies to leverage these capabilities programmatically.

Pricing and Future Plans

«Radar» is priced at around $29 per month per user, with a company plan at $399 per month that includes 20 seats, while API users are subject to custom pricing based on their needs.

In the future, the company plans to expand the service beyond podcasts to include other forms of audio, such as «YouTube» clips and newsletters, in a move that reflects a broader ambition to convert all audio content into data that can be understood and analyzed automatically.

✦ بقلم فريق دروب أيديا