New episodes now show up in minutes
Hello! Here’s what’s 🆕 since our last update:
New episode sync now close to real-time: New episodes shouldn't take a day to show up, so we made our system smarter about catching them. It now learns each show's real release rhythm (the specific days and hours a program tends to publish) and checks in most closely right when a new episode is likely to drop. That lets us keep more than 100,000 podcasts in sync every day (about 20,000 more than there are seconds in a day) while surfacing new episodes within minutes, rather than the hours or days that might be common elsewhere.
Universal entity search: New
/v1/entities/searchreturns a single ranked feed across people, companies, and topics, with nickname aliases baked in, so "Zuck" finds Mark Zuckerberg and "JLo" finds Jennifer Lopez. Each hit comes back with disambiguating context (company domain, description, popularity signals), so when you're picking an entity, you can tell the right "Apple" or the right "Jason" at a glance. Available in REST and via MCP.Cleaner corpus: We now run automated AI/content-farm detection at the publisher level, and blocklist matches from ingestion, so trending lists, search results, and alerts don't get polluted by low-quality shows.
Sharper mention matching: Entity mentions now pass an added relevance check, so look-alike names and passing brand references stay out of your results, and the mentions you see are the ones that genuinely matter.
FYI - Using the right search endpoint: A quick reminder to point your searches at the right place. Use
/v1/podcasts/searchto find shows and/v1/podcasts/episodes/searchto search within episodes. The docs now spell out the distinction more clearly, and content-search parameters likesemantic_search(which only apply to episode search) now return a clear 422 pointing you to the right endpoint if they're sent to show search, instead of being quietly ignored.Alerts get API: The API/MCP support we promised alongside last week's Alerts launch is now live. Create and manage them entirely through our REST API and MCP tools, so an agent can set them up for you.
Best,
Team Particle