Podscan.fm

Podscan.fm is the brand-monitoring tool for podcasts: near real-time transcripts of millions of episodes, with alerts when your keywords show up. The API and MCP server make it surprisingly useful for AI agents. Accuracy varies on accented or technical episodes, and heavy API usage gets expensive.

Podscan.fm built itself in the open as a one-person project, with founder Arvid Kahl narrating the build on social as it grew, and has turned into one of the more interesting tools in the podcast data layer. The core proposition is straightforward: transcribe a very large slice of the active podcast catalogue in near-real-time, and then let users get alerts when names, brands, or topics show up in episodes anywhere across that catalogue. For PR teams trying to monitor brand mentions across audio, that turns a previously invisible category of media coverage into something monitorable and reportable. For competitive intelligence teams tracking what rivals are saying or being said about, the same capability applies. The API has become a genuine strength rather than a footnote, especially as AI agents and LLM-based assistants increasingly want podcast transcripts as input for downstream tasks like research, summarisation, and content analysis. The MCP server makes that integration cleaner than it has any right to be for a project this size, and the documentation has gotten progressively better over time. The honest limits are worth noting upfront. Transcription quality varies more on accented English or technical episodes than on standard US news shows or interview podcasts, which can produce false negatives on keyword alerts when speakers pronounce brand names unexpectedly. The API pricing scales steeply for high-volume monitoring use cases, so a brand-monitoring agency tracking thousands of keywords will see real bills. Indie creators or hobbyists are usually better served by a Listen Notes plan; brands tracking podcast share-of-voice are the right buyer.