“Verbatim

Is your market data written by AI? Podcast evidence checked for human speakers

Verbatim Team, from The Art of Service, gives teams living market data from 900+ business podcasts, with a Human-Voice Check on every show. After each daily refresh every show is screened for signs it is machine-made; flagged shows wait for a person's review, and shows judged machine-made are removed from search, alerts, the API and MCP. Verbatim Team is USD 299 a month for 5 seats, with full archive search, REST API and MCP.

Start Verbatim Team See the library standard

Updated

The screen right now

Counted live each time this page loads. The screen runs after every daily refresh.

930shows screened
6flagged, held from briefs, awaiting review
2on watch, not held
0excluded after review
1cleared after review
16%of moments from a confirmed speaker

See the Excluded Shows Register · Report a show

This matters if

a detection company's own error rate
“Pangram 4's false positive rate is about 1 in 24,000”
♫
Said on Unit Economics
Hard number 1 in 24,000 Share this moment

Framework

The Voice Gate

It shows shows flowing toward a waveform-shaped slot: moments spoken by people pass into your Team workspace, and machine-made items drop to a tray.

The Voice GateIt shows shows flowing toward a waveform-shaped slot: moments spoken by people pass into your Team workspace, and machine-made items drop to a tray. Shows: 900+ business and technology podcasts, refreshed daily. Voice Gate: screened, then reviewed. Your AI: search, alerts, REST API and MCP for 5 seats. Timestamp: every moment linked to the second it was said. Excluded tray: shows a person judged machine-made, listed publicly.Shows: 900+ business and technology podcasts, refreshed dailyShows900+ business andtechnology podcasts,refreshed dailyVoice Gate: screened, then reviewedVoice Gatescreened, then reviewedYour AI: search, alerts, REST API and MCP for 5 seatsYour AIsearch, alerts, REST API and MCP for 5 seatsTimestamp: every moment linked to the second it was saidTimestampevery moment linked to the second it was saidExcluded tray: shows a person judged machine-made, listed publiclyExcluded trayshows a person judged machine-made, listedpubliclyThe Voice GateIt shows shows flowing toward a waveform-shaped slot: moments spoken by people pass into your Team workspace, and machine-made items drop to a tray. Shows: 900+ business and technology podcasts, refreshed daily. Voice Gate: screened, then reviewed. Your AI: search, alerts, REST API and MCP for 5 seats. Timestamp: every moment linked to the second it was said. Excluded tray: shows a person judged machine-made, listed publicly.Shows: 900+ business and technology podcasts, refreshed dailyShows900+ business and technologypodcasts, refreshed dailyVoice Gate: screened, then reviewedVoice Gatescreened, then reviewedYour AI: search, alerts, REST API and MCP for 5 seatsYour AIsearch, alerts, REST API and MCPfor 5 seatsTimestamp: every moment linked to the second it was saidTimestampevery moment linked to the secondit was saidExcluded tray: shows a person judged machine-made, listed publiclyExcluded trayshows a person judgedmachine-made, listed publicly
Every show meets the gate after each refresh; only what a person reviewed as machine-made drops to the tray.

What does the Human-Voice Check look at?

Six signals the library already holds for every show:

  1. whether any speaker is introduced by name in the first minutes of an episode
  2. how many moments come from named operators
  3. whether episode titles are only a date
  4. a daily or faster cadence with no named guests
  5. one brand running three or more look-alike shows
  6. formulaic episode titles

A score never excludes a show by itself. Flagged shows are held out of Advisory briefs and wait for a person. The live counts are at https://verbatim.theartofservice.com/library-standard.

Signs your market data may be machine-made

Frequently asked questions

How do I know a podcast quote came from a real person?

In Verbatim, press play: every quote links to the exact second it was said. Verbatim also marks a speaker as confirmed only when they are introduced by name early in the episode, and the Human-Voice Check screens every show for signs it is machine-made. Verbatim Team brings that same checked library into your own AI.

Are AI-generated podcasts in market research data?

They can be. Some networks now publish synthetic episodes cheaply and at volume. Verbatim screens every show after each daily refresh, holds flagged shows out of Advisory briefs, and removes shows a person judges machine-made from search, alerts, the API and MCP. Verbatim Team inherits that screen.

What is the Human-Voice Check?

The Human-Voice Check is the library screen Verbatim runs after every daily refresh. It scores each show on six signals, such as no named speakers and date-only titles. A score never excludes a show alone: a person reviews every flagged show, and only that verdict removes it. It applies to every Verbatim plan.

Does Team work with my own AI?

Yes. Verbatim Team includes a REST API and an MCP server, so MCP clients such as Claude and ChatGPT that accept a remote server URL and header can search the library directly. Results carry the quote, speaker, show, date and a link to the second, and excluded shows never appear.

What does it cost?

Verbatim Team is USD 299 a month for 5 seats, or USD 3,588 a year billed once. It includes full archive search, shared watchlists, REST API and MCP, and everything in Pro. Cancel any time; access continues to the end of the period you paid for. More than 5 seats is on request.

Can I hear every source?

Yes. Every moment in Verbatim, including every result returned to your AI through Verbatim Team, links to the exact second it was said and plays from that second. Quotes are short; the full episode belongs to its publisher, and shows can ask to be removed at any time.

Living data your AI can trust

5 seats, the full archive, API and MCP, and every source checked for a human voice.

Start Verbatim Team

Related