Skip to main content
A podcast line is only as useful as the person who said it. An analyst weighing what was said about a drug launch needs to know whether it came from a practicing cardiologist or from a host reading the headlines. A diligence team vetting a commentator needs to know whether the credentials hold up. A fund tracking a debate wants the economists and the industry analysts, not the panel of generalists. Particle reads the people who carry a podcast’s conversation, its hosts, guests, panelists and correspondents, and gives each one an expertise profile: their occupations, their fields of knowledge, how established their standing is, how senior they are, whether they still practice, and the credential that establishes it, quoted from the public record. Every guest appearance is read too: was the guest speaking within their field, was the appearance mainly promotional, and how did the show introduce them? Those labels then work everywhere: as filters on episode lists, transcript search and mentions, as rankings of who is being booked, and as a split of who is talking about a company. And wherever a response names a speaker or a guest, it says what they know, so you can tell who said a line without looking anyone up.

​
For AI agents

When a question restricts who is speaking by occupation, profession, credentials or industry (“doctors”, “medical professionals”, “economists”, “people in finance”, “lawyers”, “analysts”), filter by expertise codes. Never rely on search wording.
  1. Resolve the words to codes first. Call particle_expertise_resolve (free) with the words the question uses. Over REST, use GET /v1/people/occupations?q=… or GET /v1/people/fields?q=…. Pass parent to see a group’s children. A broad code covers every code below it.
  2. Filter by the code. Pass the code or slug as guest_occupation or guest_field on search, mentions, episode lists and the timeseries. On the guest directory and particle_expertise_get, use occupation or field. When credentials matter, add guest_standing=established or guest_practicing=true. Never put the profession in semantic_search or keyword_search to stand in for who is speaking.
  3. Retry on unresolved_reference; don’t drop the filter. The error names the parameter and suggests close codes. Resolve again and retry with a code it returns. A search without the filter answers a different question.
  4. Check each speaker before you attribute a line.
    • Search, mentions and episode speakers list each identified speaker with their compact expertise: occupation codes, standing and seniority. In MCP that is a **Speaker:** Name (slug, ROLE) — Title (code) · standing · seniority row.
    • The guest_ filters pick episodes, and on search segments, in which such a guest speaks. So results also carry host lines.
    • Timeseries counts with a guest_ filter count episodes or keyword segments, not lines the experts spoke. Don’t present them as what the experts said.
For “What risk factors do medical professionals talk about with GLP-1s?”:
  1. Resolve “physicians” or “cardiologists” to a code.
  2. Search semantic_search="risks and side effects of GLP-1 drugs" with guest_occupation=<code> and guest_standing=established.
  3. Quote only the lines whose speaker’s expertise carries that code.

​
What you can build

  • Expert sourcing. Find practicing cardiologists, established energy analysts or distinguished economists, with the credential that backs each one. Use it for call lists, panels and expert-network outreach.
  • Signal by source. Restrict transcript search, mentions and their trend lines to what credentialed people said. Hear what equity analysts say about Nvidia, apart from host chatter, or what cardiologists say about GLP-1 drugs.
  • Commentator diligence. Check a speaker’s standing, seniority and experience, and read the credential, before you cite them or book them.
  • Who is shaping a debate. Rank the economists booked most this month, and the shows that book the most financial analysts.
  • Share of voice. Split the conversation about a company or a topic by the professions doing the talking.

​
Choosing the right endpoint

​
The data

​
A person’s expertise

Profiles are built from the public record: biographies, professional profiles, encyclopedia entries and how shows introduce the person. A code is listed only when we’re confident it applies, so a profile names what someone demonstrably does rather than everything they’ve touched. Fields the record doesn’t support, such as organization or career_start_year, are left out.

​
A guest appearance

When we aren’t confident that the person linked to an appearance is really the person speaking, that appearance is left out of every expertise filter until the link is settled. A namesake’s appearances don’t count toward someone’s record.

​
Beside every speaker

Search matches and mention episodes carry speakers: the identified people speaking in their lines, matched to each line by name. Episode speakers (GET /v1/podcasts/episodes/{id} and /speakers) and guest listings (GET /v1/podcasts/guests, a show’s roster and trends) carry the same compact expertise on each person:
It lists up to three of the most specific SOC 2018 occupations and ANZSRC 2020 fields, never a code beside its own parent, each ready to pass to guest_occupation or guest_field. expertise is omitted when the person has no profile yet, when their link is unsettled, when that appearance may describe someone else, or when the speaker is in a role profiles don’t cover, such as a moderator or a narrator. Hosts, guests, panelists and correspondents carry theirs. The full profile is GET /v1/people/{id}/expertise.

​
A person’s profile

GET /v1/people/{id}/expertise returns one person’s profile. {id} is a person’s slug or ID.

​
Find experts

GET /v1/people/expertise ranks the people listed under an occupation or a field, most likely first, so every request names an occupation or a field (one without either returns a 422). Narrow it with any of the other filters: Practicing cardiologists with established standing:
The first results include Martha Gulati, Director of Preventive Cardiology at Cedars-Sinai; Rachel Bond, System Director of Women’s Heart Health; and Faraz Ahmad, Associate Director of AI at Northwestern Medicine’s Bluhm Cardiovascular Institute. Each is someone who has spoken on a podcast, so you can follow them straight into their appearances and what they said. Established financial and investment analysts (occupation=financial-and-investment-analysts&standing=established) start with sector specialists: Bloomberg Intelligence’s analysts covering EMEA media and telecom, and global fertilizer markets, and a Wells Fargo equity analyst covering integrated oils and refiners. Distinguished economists (occupation=economists&seniority=distinguished) start with Eric Hanushek of the Hoover Institution, Thomas Sowell, the Brookings Institution’s Martin Baily, a former chair of the Council of Economic Advisers, and Alicia Munnell of Boston College’s Center for Retirement Research.

​
Occupations and fields

Every occupation and field parameter accepts a code, a slug or a title, such as 19-3011, economists or Economists. Browse the standards to find the right one:
  • GET /v1/people/occupations and GET /v1/people/occupations/{code} walk SOC 2018, from major groups such as 13-0000 Business and Financial Operations Occupations down to detailed occupations such as 13-2051 Financial and Investment Analysts, with the ISCO-08 equivalents of each.
  • GET /v1/people/fields and GET /v1/people/fields/{code} walk ANZSRC 2020, from divisions such as 38 Economics down to groups and fields such as 3801 Applied economics.

​
Who is being booked

GET /v1/people/occupations/{code}/guests ranks the people in an occupation by their guest appearances in a window of publication dates (the last 30 days by default, up to 90), then by how many distinct shows booked them. Raise min_podcasts to favor people booked across the medium over one show’s regulars. GET /v1/people/fields/{code}/guests does the same for a field of research.
The inverse, GET /v1/people/occupations/{code}/podcasts, ranks the shows that book an occupation. Over the last 30 days, the shows booking the most distinct financial and investment analysts were Schwab Network (148 analysts), CNBC’s Closing Bell (85) and Bloomberg Surveillance (77). That’s where to listen for sell-side and buy-side views in volume. Use min_guests to keep only shows that book several people from the occupation.

​
Episodes with expert guests

The episode list takes guest filters: guest_occupation, guest_field, guest_standing, guest_seniority, guest_practicing and guest_in_field. They select episodes with at least one guest who matches all of them, so guest_in_field=true with an occupation keeps episodes where that kind of expert spoke within their field. Like practicing, guest_practicing needs a detailed SOC 2018 guest_occupation.
Recent results include Yale’s Health & Veritas on “The Revolution in Heart Care with Eric Velazquez” and a preventive-cardiology episode on exercise, nutrition and longevity. The same filters work on GET /v1/podcasts/episodes/timeseries to chart how often such episodes appear. The timeseries needs a subject to count, and guest_occupation or guest_field serves as one; guest_standing, guest_seniority, guest_practicing and guest_in_field narrow a subject rather than replace it.

​
What experts said about a topic

Transcript search takes the same guest filters, and with them it keeps the segments in which a matching guest speaks, its preview opening on that guest’s first line. A segment can also carry the host’s lines, so read each line’s speaker in the match’s speakers. Ask how tariffs affect prices, and hear from established economists:
The top result is NPR’s Planet Money, “Days of our Tariffs”, with Harvard Business School’s Alberto Cavallo:
“In principle, you know, tariffs are a tax on imported goods, so why are domestic goods also increasing? But there are very valid reasons for that… It could be that some of these domestic manufacturers have imports in their inputs.”
The next is the Free To Choose Media podcast, where Jeff Ferry cites the steel study that found “the 25 percent tariff on steel raised… the price paid in the U.S. for steel by two point four percent, a tenth of the headline value of the tariffs.” Swap in guest_occupation=cardiologists and ask about GLP-1 drugs and heart risk, and the results come from physician shows: The Podcast by KevinMD on GLP-1s and the inflammation tests a patient needs, and the ISTH podcast on GLP-1s and blood clots.

​
What experts said about a company

Mentions take the guest filters too, and keep only the lines that a matching guest spoke. Every line in which a financial or investment analyst named Nvidia:
On Schwab Network, Bob Lang reads the options market: “always have a lot of… positive bullish flow in Nvidia all the way out into January and February of 2027… those strikes are all the way up to 240 to 250, so somebody’s looking for some large, large moves in that stock.” Each line comes with the speaker, the segment and the time in the episode, so you can cite it or clip it. GET /v1/podcasts/mentions/timeseries takes the same filters and counts those lines per day, week or month: how often analysts bring up a company, apart from everyone else.

​
Share of voice

GET /v1/podcasts/mentions/share-of-voice splits the lines that mention a company, person or topic by the occupations or fields of the people who said them. Use it to see which professions are driving the conversation about a stock, a drug or a policy.
Over those 30 days, guests with an expertise profile spoke 3,431 of the 14,760 lines that named Nvidia. Financial and investment analysts led with a fifth of them, chief executives followed, then journalists. Pass a group’s code as guest_occupation to mentions to read what that group said. mentions counts every line in the window. attributed_mentions counts the lines spoken by someone with an expertise profile, and each group’s share is of those. A person listed under two occupations counts in both, so shares can add up to more than 1.

​
An episode’s guests

GET /v1/podcasts/episodes/{id}/expertise lists each guest on an episode with their profile and how the appearance reads:

​
A show’s guests

GET /v1/podcasts/{id}/guests/expertise profiles the guests a show books, across every analyzed guest appearance: how many were by established, self-described and unverified guests, how many stayed within the guest’s field, how many were mainly promotional, the seniority mix, and the occupations and fields booked most.
Each share is of guest_appearances. A guest listed under two occupations counts in both, so occupation shares can add up to more than 1. The podcast list filters on the same profiles. guest_occupation and guest_field keep shows that have booked such a guest; min_established_guest_share, min_in_field_guest_share and max_promotional_guest_share bound the shares above, and min_analyzed_guest_appearances leaves out shows with too few analyzed appearances for their shares to mean much. Shows that book established economists:
A show with no analyzed guest appearance has no profile, so it never matches guest_occupation, guest_field, a share bound above 0 or max_promotional_guest_share. A minimum share of 0 sets no bound.

​
The guest directory

The guest directory, GET /v1/podcasts/guests, takes occupation, field, standing, seniority, practicing, capacity and employer, so you can browse the guests of a profession by their lifetime appearances. A guest’s appearances, GET /v1/podcasts/guests/{id}/appearances, take in_field=true and exclude_promotional=true to keep the episodes where they spoke as an expert rather than to sell something.

​
Things to know

  • Hosts have profiles too. Anyone who has appeared as a host, guest, panelist or correspondent gets an expertise profile, so speakers=all on share of voice includes hosts. Lines from speakers without one, such as narrators, announcers and soundbites, count toward mentions but not attributed_mentions. Appearance reads (in_field, promotional, introduced_as) and the guest_ filters cover guests and panelists.
  • Profiles follow the record. When a person’s public record changes, such as a new role or new published work, their profile is updated.
  • Codes overlap by design. A person can be listed under several occupations and fields, such as an economist who is also an author, and counts under each one.