Podcasts are the last frontier of authentic expertise.
We make them searchable.
Millions of hours of expert conversation are locked in audio. Founders sharing hard-won lessons. Industry leaders debating where their market is heading. Investors revealing what they're watching. Executives explaining strategy in their own words.
While the rest of the web fills with AI-generated content, podcasts remain real people having real conversations. Unfiltered, unrehearsed, and often more revealing than any press release or social media post.
But who has time to listen to everything?
That's why StationLM exists.
StationLM is the knowledge graph for human speech. It indexes hundreds of thousands of podcast episodes, transcribes and AI-analyzes them, and pulls out the people, companies, topics, quotes, and insights worth knowing — so you can search what was actually said without listening to a minute of audio.
It's free and open. Browse and search everything without an account, or ask the StationLM Agent in plain English. Developers and AI agents can query the same knowledge graph through the StationLM API, so ChatGPT and Claude can cite what was said with sources. The only thing that costs anything is analyzing a brand-new episode on demand — priced at roughly what it costs to run.
700,000+
Episodes Indexed
1,900+
Shows Analyzed
20+
Categories
Built by The London Bureau.
StationLM started from a simple frustration: there was no way to search what people actually said on podcasts. Google doesn't index audio. Social media only captures soundbites. And nobody has time to listen to everything.
So we built the tool we wished existed — and kept it running because it's a working demonstration of what we do at The London Bureau: build AI systems that turn messy real-world data into signals you can act on. If you have a corpus of your own worth mining, that's the kind of problem we like.
See what people are actually saying.
Free and open. No account needed to browse.