Podcasts have always been a black hole for search. You can Google a blog post from 2011, but the sharpest thing a founder said on a two-hour show last Tuesday? Gone the moment the episode ends, buried in audio no crawler can read. Particle wants to fix that.
The startup this week launched Radar, a podcast intelligence platform that has transcribed and analyzed more than 130,000 podcasts. Its pitch is simple and, honestly, overdue. Take the vast library of spoken conversation that fuels modern media, turn it into text a machine can parse, and make all of it searchable on the open web. Then hand the same firehose to AI agents through an API and an MCP connection.
Turning audio into an index
Consider how much knowledge lives inside podcasts and never surfaces anywhere else. Interviews with researchers, off-the-cuff product roadmaps, arguments between economists, a comedian’s aside that turns out to be a real news tip. None of it shows up when you search, because search engines index words on pages, not sound waves moving through your earbuds.
Radar’s answer is to transcribe at scale and then analyze what those transcripts contain. With more than 130,000 shows processed, the platform is building the kind of index that treats a spoken sentence the way Google has always treated a written one. A quote becomes findable. A claim becomes traceable to the episode and moment it was made. Conversations that used to vanish into the archive become a queryable record.
That alone would be useful. What makes Radar interesting is who Particle expects to actually use it.
Built for agents, not just people
Most search products are designed for a human squinting at a results page. Radar is built with a different reader in mind. Through its API and MCP, the platform exposes podcast content directly to AI agents, the software assistants that increasingly go looking for information on their own rather than waiting for a person to type a query.
MCP, the Model Context Protocol, is the plumbing that lets these agents plug into outside data sources in a standard way. By supporting it, Particle is essentially saying that podcasts should be a first-class source for AI the same way news articles and databases already are. An agent researching a company could pull what its CEO said on three different shows. A model summarizing a debate could cite the actual audio it came from, not a secondhand recap.
This is a meaningful shift in how spoken media gets used. For years the value of a podcast conversation stayed locked inside the listening experience. You had to be there, ears on, for two hours. Radar treats that same conversation as structured data other software can act on, which changes what a podcast is worth long after the recording stops.
Why the timing matters
Particle is stepping into a moment when AI systems are starving for high-quality, human material to draw on. Text from the open web has been scraped, argued over, and in some cases walled off behind licensing fights. Audio, by contrast, remains a huge and largely untapped reserve of genuine human conversation. Making 130,000 podcasts legible to machines opens a door that has mostly stayed shut.
There are real questions hanging over all of this, and they are worth taking seriously. Podcasters may not love the idea of their words becoming raw material for AI agents they never agreed to feed. Rights, attribution, and consent get complicated fast when a platform sits between the person who spoke and the machine that quotes them. How Particle handles those tensions will say a lot about whether Radar becomes infrastructure the industry trusts or a flashpoint it resents.
Accuracy is the other test. Transcription at this scale is hard, and a search index is only as good as the text underneath it. A misheard name or a garbled figure does not just produce a bad result for a human reader. It gets passed downstream to an agent that may treat it as fact and repeat it somewhere else.
Still, the direction is clear. Search is no longer only about pages, and increasingly it is not only about people doing the searching. Radar is a bet that the next great index will include the things we say out loud, and that the readers hungriest for it will be the machines. Watch whether the podcast industry decides to open the door or push back against it.
For more coverage of AI and podcasting, visit Mylistingo.
Source: Original Article
The Netherlands, for internationals
One email a week: the week's news for internationals in the Netherlands, new practical guides and what changed in the cities. No spam, unsubscribe any time.
