A content engine for CoreVitals

AI agents do the research, drafts and checks. People do the judgment. Your clinicians only see the claims that actually need them.

Workflow logic by Aziz Hamidov, built around the tools CoreVitals already runs.

Your stack
Mine
Who does it

Tap a tool to see where it plugs in. Tap a step to open it.

The weekly loop

Seven steps. Five run on their own, two need a person. Every Monday the results feed the next round.

1
AI agent

Find topics that make money

Starts from what converts, not from search volume.

Search ConsoleGA4Google AdsAhrefs MCPn8n
Pulls
  • Queries and pages from Search Console
  • Membership signups and MRI bookings from GA4
  • Google Ads search terms that already convert
  • Ahrefs keyword gaps vs Thriva, Medichecks, Randox
Does
Scores every topic on demand, buying intent and product fit (blood membership or London MRI). Drops topics GetScanned already owns, like "book a scan", prices and "near me".
Gives
A ranked topic list, refreshed weekly.

Your Google Ads data already shows which searches buy. That's free validation, so SEO starts there.

2
AI agent

Brief from Google and AI answers

Reads what ranks and what ChatGPT says, then finds the gap.

Ahrefs MCPClaude Code
Pulls
Top 10 Google results, plus what ChatGPT, Gemini and Perplexity answer and which sites they cite.
Does
Spots questions nobody answers well, missing data and weak angles. Builds the brief: angle, headings, FAQs, internal links and the right CTA (blood test or MRI).
Gives
A one page brief per topic.

Health searchers open 3 or 4 tabs before they buy. The brief makes sure ours also answers the next question they'd ask.

3
AI agent

Draft from approved sources only

Every medical sentence comes with its source attached.

Claude Code
Uses
The brief, a CoreVitals tone guide and a fixed source library: NHS, NICE, PubMed and your clinical team's own notes.
Does
Writes the first draft with inline citations. Shows how that result looks inside the CoreVitals app, colour coded and in plain English, so the product is part of the answer.
Gives
A cited draft, ready for checking.

A paper or PDF report vs an app that explains your results is the strongest selling point. Every article should show it.

4
AI agent

Claim and compliance check

Same colour logic as your app:

Claude Code
Does
  • Pulls out every health claim and checks it against its source
  • Green: matches. Amber: overstated or weak source. Red: no source, or diagnostic wording
  • Rewrites risky lines. CoreVitals is not a diagnostic service, so no "detects" or "diagnoses"
  • Checks claims against ASA / CAP code rules and adds disclaimers
Gives
A claim report like the one at the top of this page.

This is the step that makes clinician review fast instead of a bottleneck.

5
Human gate

Two people sign off

Me for the writing, a clinician for the medicine.

EditorClinician
Gate 1
I edit for angle, readability and real examples, and cut the AI fluff.
Gate 2
A clinician reviews only the amber and red claims, then signs off. Their name goes on the page as medical reviewer.
Gives
An approved article.

Clinician time is the most expensive part of health content. The goal is minutes per article, not hours.

6
AI agent, human clicks publish

Publish to Webflow, reuse in Klaviyo

One article becomes a page, a test and three emails.

WebflowOptibaseKlaviyon8n
Webflow
Creates the CMS item as a draft through the API, never straight to live. Adds MedicalWebPage, FAQ and reviewedBy schema (right now only Organization schema shows on the site) and internal links to the test pages.
Optibase
A/B tests the title and CTA, like "Book your blood test" vs "See what 100+ biomarkers show".
Klaviyo
Turns the article into email blocks for existing flows: before your blood draw, understanding your results, time for your 6 month retest.

Content that also feeds retention gets paid for twice. The 6 month retest is where membership revenue lives.

7
AI agent

Measure, refresh, repeat

A Monday report that also writes next week's to-do list.

Search ConsoleGA4ClarityKlaviyon8n
Checks
  • Rankings and clicks in Search Console
  • Signups and bookings per page in GA4
  • Scroll depth and rage clicks in Clarity
  • Which readers became members, via Klaviyo and UTMs
  • A fixed prompt list run in ChatGPT, Gemini and Perplexity, logging mentions and citations
Does
Pages that are slipping go to a refresh queue. Winners get spin-off topics.
Gives
The weekly report, plus new input for step 1.

Most content teams only publish. Refreshing what already ranks is usually the cheapest traffic there is.

Back to step 1, every Monday

What I'd track

Organic membershipsBlood test signups from organic and AI search
MRI bookingsLondon full body MRI bookings from content
AI citationsMentions and links in ChatGPT, Gemini, Perplexity
Clinician minutesReview time per article, should keep dropping
Topic to liveDays from step 1 to a published page