Particle Podcast Intelligence Blog

Hello!

Here's what's new on the Particle Podcast Intelligence API over the last few weeks.

Related shows, related episodes, and recommendations

Our biggest addition this month: five new endpoints:

• GET /v1/podcasts/{id}/related returns the shows that are most similar to a given podcast. Each one carries a score plus a simpler band you can branch on: strong, moderate or weak. Add include=basis to see why the two were paired, whether that is similar recent episodes, overlapping topics, guests they share (named), the same publisher, the same sponsors, or the same categories. For just a taste, include=related on podcast detail returns the top five.

• GET /v1/podcasts/episodes/{id}/related answers "who else covered this": episodes from other shows on the same story or subject. published_within_days narrows it to a recent window for news work, and same_podcast=true lets the show's own episodes back in.

• GET /v1/podcasts/{id}/recommendations/guests is your booking pipeline: people who have been guests on shows related to yours but never on yours. include=via shows which shows booked them.

• GET /v1/podcasts/guests/{id}/recommendations/podcasts is the pitch list: shows a person has not appeared on yet, ranked by how close they are to the ones that booked them as a guest or panelist.

• GET /v1/podcasts/{id}/recommendations/sponsors🔸 is the prospecting list for a show selling its own inventory: sponsors that advertise on shows related to yours but not on yours, ranked by how related those shows are, how much the sponsor buys there, and how recently. include=contacts adds up to three people at each sponsor's company most likely to own the sponsorship decision, each with a title, a role, and a LinkedIn URL where we have one. exclude_top_advertisers=N hides the biggest advertisers so you can see the long tail.

Related shows, related episodes, and recommendations

• Related shows have to clear a minimum relatedness score, so a show with too few analyzed episodes returns a short or empty list instead of fifty barely-related ones.

• When a publisher runs several feeds for the same show, that show appears once, and those feeds count as a single venue in the recommendations. A news outlet's own correspondents no longer outrank guests that several independent shows chose.

• An empty first page carries a coverage note (not_computed, none_above_floor, or filtered_out), so you can tell "we have not worked it out yet" from "there is nothing" from "your filters ruled it out".

• GET /v1/companies/{id}/podcast/recommendations🔸 answers where a company should advertise next. It takes every show the company has never bought, then ranks them by how closely they relate to the shows its sponsors already run on, weighted by how much it spends there. Tested against the buys advertisers actually went on to make, it found them at 13 times the rate of picking by category and popularity. include=via shows the company's own shows behind each pick, and you can filter with min_score and language. On MCP: include: ["podcast_recommendations"] on particle_company_get.

Look up an episode with whatever ID you already have

GET /v1/podcasts/episodes/lookup is a 🆕 endpoint that allows you to look up episodes using eight ID types. Send up to 100 at a time, and you get one result back for each.

• guid, the ID an episode carries in its publisher's own RSS feed. This is the one most podcast tools key on, and at roughly 89% of episodes it covers more ground than anything else here.

• podcastindex, a PodcastIndex episode ID. These are numeric and always unique, so there is never any question which episode you get.

• youtube, matched against the videos we have already found for an episode. Send the bare video ID or a full URL: watch?v=, youtu.be, /live/, /shorts/ and /embed/ all work.

• apple (or itunes), the ?i= value in an Apple Podcasts URL.

• megaphone, omny, acast, art19, the hosting platform's own episode ID, which we read out of the episode's audio URL.

Trends is readable from the API

Trends, the Radar surface that opened at the start of September, is now available through the API. We call them entity charts: a daily ranking of the people, shows, and things podcasts talked about the most. Think of a music chart where artists are entities, and the airplay is podcast conversation.

• GET /v1/podcasts/entities/charts lists the most up-to-date version of every chart.

• GET /v1/podcasts/entities/charts/{category} returns a single chart, up to 100 entries deep. Categories are all, tv-series, movies, video-games, and guests. New editions are published daily, covering either 7d or 24h (all is 7d only).

Each entry tells you:

• Where it sits, and where it was. rank, previous_rank, movement (new, up, down, flat), and rank_change. Editions are daily, so on a 7d chart you are comparing this rolling week against yesterday's, not against last week.

• The counts behind it. podcast_count is how many separate shows mentioned it. episode_count and mention_count are there for context but never affect the ranking, since one recap network re-airing a single clip can inflate them. You get all three for the previous window too.

• Momentum. signals.state sums the entry up as breaking, rising, steady, settling, cooling or new, with the underlying averages available if you want them.

• Acceleration. accel_multiple compares how fast coverage grew on the last full day against that entity's own six-day average, adjusted for how much gets published on any given day. Add sort=acceleration to reorder the whole chart by it, which surfaces something coming from nowhere without it having to beat the long-running favorites.

• Why it moved, with proof. why gives you a one-line explanation when we have one, the quote behind it, the episode it came from, and the timestamps to play it. Only show or play the quote when verifiable is true. clip points to an existing highlight when one contains the quote.

• Optional extras. include=timeseries adds daily history for each entry. include=driven_by points to the entry driving a related one, so you can group something like a tribute song under the artist.

The guests chart works a little differently. It ranks people by how many separate shows they appeared on, and each entry comes with that guest's recent appearances and their best clip from the window.

Keeping AI-generated shows out of your results

We can now spot AI-generated podcasts (synthetic voice and script, text read by a synthetic voice, syndication spam, brand impersonation) and remove them. They are excluded from related shows, related episodes, and every recommendation.

Pay per request with x402, no account required

You can now pay for a call without holding an API key. Every billable /v1/* endpoint and POST /mcp accepts an x402 USDC micropayment on Base.

The flow: send a request with no key, and you get back a 402 with the price in a PAYMENT-REQUIRED header. Your agent signs the payment, retries with a PAYMENT-SIGNATURE header, and the response comes back with a receipt in PAYMENT-RESPONSE. If a request fails, you are not charged.

Every price, per endpoint and per tool: api.particle.pro/llms.txt. How it all works: docs.particle.pro/x402. The discovery manifest is at /.well-known/x402, and the public OpenAPI document now answers at api.particle.pro/openapi.json with payment terms attached to every payable operation.

One change to check: alerts now watch a single entity

particle_alert_create, particle_alert_update and particle_alert_preview, along with their REST equivalents, now accept exactly one entity. Send more, and you get an invalid_parameter error asking you to create one alert per entity. This update patches a bug that allowed some users to create alerts using multiple entities. Alerts you set up with multiple entities (before this change) will continue to work, keep watching their full list, and keep delivering as normal, but the next time you edit their entities field, you will need to narrow it to one. This also applies to all new alerts you create.

The running record is at docs.particle.pro/changelog and docs.particle.pro/mcp/changelog (MCP). Reach out with questions or feedback: api@particle.pro.

🔸Indicates Premium features that are only available on Business & Enterprise plans. You can upgrade to the Business plan via your account Settings > Manage plan. Email api@particle.pro for more information on Enterprise plans.