← The AI Hype Audit — all 256 verdicts

PARTLY "Claude can build this, but it's never gonna design this" — a fair point wearing an unfair word: 'never' Facebook · Aug 12, 2026

The claimTikTok, Sep 2 2026, 56 seconds (6,741 plays / 276 likes / 18 comments / 280 saves / 89 shares). Caption: "Claude Fable 5.1 + Public databases are an elite combo. (I should be charging for this) #claude #medclinic #leadgen". Title card: "Claude Fable 5.1 + Public Databases = infinite lead glitch". On screen: a Netlify dashboard, "THE MEDICAL CLINIC LEAD MACHINE", showing "5,935 BRAND-NEW CLINICS IN THE TRAY - POURED FROM 105,139 FEDERAL RECORDS", "the 3 weekly files CMS has published since the last monthly cut", "NEWEST WEEK ALONE: 2,048 CLINICS", a banner for "BILLING, CREDENTIALING, MALPRACTICE, STAFFING, EQUIPMENT" vendors, and a sources page stating "COST $0 - no API keys, no scraping, public files". In replies the creator confirms "the npi registry lookup" ("Yup, light work."), that it "all runs in a cloud server with no need for a UI", that it returns phone numbers ("Yup"), and that he runs "5 of 20x plans". A commenter adds "I have this exact set up but I've been using Apollo".

The build is more honest than the hook. On screen it is a real dashboard that reads the three NPPES weekly files (Aug 3-23 2026, 105,139 records), keeps the 5,935 brand-new Type 2 organization NPIs, tags each with the NUCC taxonomy (883 codes), and grades leads on shown math: phone on record, named official is the decision maker, filed in the newest week. It states "COST $0 - no API keys, no scraping, public files", masks names and phones "by policy" and links to the official registry. The timing thesis is sound for billing, credentialing, malpractice, staffing and equipment vendors: a fresh organization NPI marks a clinic's setup window. The data checks out: NPPES is free and public, the API returns addresses[].telephone_number, address_1, city, state, postal_code and address_purpose, and there is no email field anywhere. The framing is the hype. CMS has published these files since the May 30 2007 Data Dissemination Notice; Definitive Healthcare (median $50,533/yr), ZoomInfo ($33,500/yr) and Apollo ($49-$119 per user per month) sit on the same rows. "5 of 20x plans" is $200 x 5 = $1,000/mo to parse three CSVs, and Anthropic's consumer terms bar sharing an account or scripting it outside the API. Nothing on screen shows revenue, and outreach still meets 47 CFR 64.1200 consent rules and CAN-SPAM at up to $53,088 per email.

What holds up

  • The source is real, free and public: NPPES FOIA-disclosable data, an open v2.1 API (live call returned addresses[].telephone_number, fax_number, address_1/2, city, state, postal_code, address_purpose and basic.authorized_official_telephone_number), a monthly full file (~1.1 GB zip) and weekly incrementals (~7 MB) at download.cms.gov/nppes, matching the dashboard's stated "COST $0" and "WINDOW 2026-08-03 to 2026-08-23 (3 weekly files)".
  • The dashboard is a competent, transparent use of that data: Type 2 organization filter, NUCC taxonomy (883 codes), shown grading math, masked names and phones with a one-click link to the official npiregistry.cms.hhs.gov record, and a defensible timing thesis (a new organization NPI precedes payer credentialing, billing-partner selection and malpractice binding, so B2B vendors reach clinics inside their setup window).
  • Claude Max 20x is a real $200/mo plan (5x is $100), so "5 of 20x plans" is a consistent $1,000/mo, and Anthropic does allow Claude Code to run headless on Pro/Max, so "cloud server with no need for a UI" is not fiction.

What doesn't

  • 'Never' is unsupportable — current Claude produces coherent, sometimes distinctive visual design today; the gap is context, not capability
  • The slick site used to 'prove' human superiority is exactly what AI produces unassisted — and viewers can't verify it wasn't AI-assisted
  • Soft self-interest: it's authority content from a designer whose livelihood the claim defends
  • The build-vs-design binary is already stale — the real 2026 split is AI defaults vs human-directed AI output, and the winners use both

The catch

The "public database" is a free weekly CMS file that healthcare data vendors have resold for nineteen years, and the dashboard on screen says so itself: $0, no API keys, no scraping, phones masked, verify at the registry. The hook sells a glitch and a paywall that the build never claims, and running it on $1,000 a month of stacked consumer Claude accounts is the one part that is both unnecessary and off-terms.

How to actually do it

  • Download the free NPPES weekly incremental files from download.cms.gov/nppes/NPI_Files.html (about 7 MB each), keep Entity Type Code 2 with an enumeration date in the last 30 days, and filter by the NUCC taxonomy codes your product actually serves; Excel, DuckDB or a 20-line script does it, and if you want Claude to write the script, a $20/mo Pro plan or pay-per-token API access is plenty.
  • Use the timing thesis, not a blast list: a new organization NPI is a clinic in its setup window, so send a single postal or manual-call introduction to the named official at the business number on record; no autodialers, SMS blasts or prerecorded voicemails without written consent, and keep an internal do-not-call list.
  • NPPES has no emails, so if you email, build the list from the clinic's own website or a paid enrichment tool and make every message CAN-SPAM compliant (real postal address, working opt-out, honest subject line); do not rely on five consumer accounts for a job the API or one Pro seat covers.

Free weekly CMS file plus a transparent dashboard is real and useful; "infinite lead glitch", "I should be charging for this" and the five-account Max stack are the hype.

Effort
A brief takes an hour; it's the highest-leverage hour in the whole build
Cost
$20/mo DIY-with-taste; a few hundred to ~$2K for AI production + human direction; $3K–15K+ for full human design
Stack
A real brief + Claude for production + a human on intent
Confidence
Medium
Posted by
a working UX designer posting opinion commentary, no product

See the original claim →

We test hype for free. We build the real thing for a living.

Thirty minutes, no pitch — and you'll leave with something useful either way.

Book a call with Todd or start with the free Business Checkup →

The Verdict Weekly

Three verdicts every Friday. Free forever, unsubscribe anytime, no spam — that would be ironic.

© Schreier Group · schreiergroup.com · See a wild AI claim? Drop it here and we'll test it.