Agentic Workforce Company
SEO Automation · AEO / GEO

I built the infrastructure to make a health brand citable by AI search.

Search is splitting in two: the ten blue links, and the answer an AI gives directly. I structured Dr. Berg Nutritionals so ChatGPT, Perplexity, Gemini, Google AI Overviews, Claude, and Copilot can find, parse, trust, and cite it — llms.txt, A2A endpoints, MedicalWebPage-grade schema, verified author and medical-reviewer markup, and a citation-tracking method to measure whether it lands. This is answer-engine optimization built as real technical infrastructure, not a content afterthought.

ShippedROLE Designer & ImplementerCLIENT Dr. Berg NutritionalsBUILT 2025 – Mar 2026
DisciplineAEO / GEO
ClientDr. Berg Nutritionals
PlatformWordPress VIP · Rank Math
Assistants6 targeted
StatusShipped · live in production
The Problem

AI answers health questions directly now.

Being absent from the answer — or cited without proper authorship signals — is both a visibility and a trust problem.

For a medical-adjacent brand, that’s not optional.

Traditional SEO optimizes for a crawler and a ranking. AI search optimizes for a model that has to decide which sources are authoritative enough to quote.

Dr. Berg’s site needed to be machine-legible to LLM crawlers, needed unambiguous entity and authorship signals — who Dr. Eric Berg is, who medically reviewed the content — and needed a way to measure citation presence across the major assistants so the work could be steered by data rather than assumption.

The Approach

Three layers: legibility, authority, measurement.

I treated AEO/GEO as a technical-SEO discipline: make the site legible to AI crawlers, make its authority explicit through structured data, and measure citation presence across the major assistants.

Machine legibility. An llms.txt file published live via Rank Math describing the Dr. Eric Berg DC entity, key topics, and entity links — plus AI-friendly crawler directives, A2A endpoints, and NLWeb for machine access.

Authority signals. sameAs identity across all pages, a clean Organization / Person entity separation, movement toward MedicalWebPage-grade markup, and a medical-reviewer audit feeding author/reviewer signals — exactly the E-E-A-T cues AI systems weigh.

Measurement. AI-citation tracking across six assistants, using prompt auditing to check whether and how the brand surfaced in AI answers.

AI search is a technical-SEO discipline, not a content afterthought. The job is to make a health source legible enough to parse and authoritative enough to quote.

— Design principle · built to be cited
Architecture & Stack

The AI-search stack.

A legibility layer exposes the site to AI crawlers. An authority layer makes the brand’s expertise explicit in structured data. A measurement layer watches whether the six major assistants actually cite it.

Fig.01 — Three-Layer AI-Search StackThree-layer AI-search infrastructure stack feeding six AI assistantsThree stacked layers on the left describe the infrastructure. The legibility layer exposes llms.txt, A2A endpoints, NLWeb, and AI-crawler directives. The authority layer holds sameAs profiles, a separated Organization and Person entity graph, movement toward MedicalWebPage schema, FAQPage coverage at 98.7 percent, and author and medical-reviewer markup. The measurement layer tracks AI citations across platforms. On the right, a panel lists six AI assistants — ChatGPT, Perplexity, Gemini, Google AI Overviews, Claude, and Copilot. Accent flows run from the legibility and authority layers to the assistants; a dashed amber path runs from the assistants back to the measurement layer, which is annotated as methodology under review.LAYER 01 · LEGIBILITYmake the site readable to AI crawlersllms.txtA2A endpointsNLWebAI-crawler directivesLAYER 02 · AUTHORITYmake expertise explicit in structured data8 sameAsOrg + PersonPerson / Org splitDr. Eric Berg DCMedicalWebPage+ FAQPage 98.7%Revieweraudit · stagedLAYER 03 · MEASUREMENTtrack whether the assistants actually cite itAI-citation / prompt audit6-platform trackingquantified output · under reviewAI ASSISTANTSChatGPTOpenAIPerplexityanswer engineGeminiGoogleAI OverviewsGoogle SearchClaudeAnthropicCopilotMicrosoft / Bing AIMEASUREMENT OBSERVES · METHODOLOGY UNDER REVIEW

The legibility layer (llms.txt, A2A, NLWeb, AI-crawler directives) makes the site machine-readable. The authority layer (8 sameAs profiles, a clean Organization/Person entity split, movement toward MedicalWebPage schema, 98.7% FAQPage coverage, and staged medical-reviewer markup) makes its expertise explicit. The measurement layer tracks citations across all six assistants. Quantified citation figures are held back pending methodology review — the proven claim is the infrastructure and the measurement approach.

Machine Legibility

llms.txt (live via Rank Math)A2A endpointsNLWebAI-friendly robots.txt

Entity / Structured Data

rank_math/json_ld filter8 sameAs · Org + PersonPerson / Org separationMedicalWebPageFAQPage

E-E-A-T / Authorship

author entity markupmedical-reviewer auditlastReviewed signaling

Platform

WordPress VIPRank Math SEO PROschema-optimization mu-plugin

Measurement

AI-citation trackingprompt auditingBing AI performance signalsGSC snapshots

Positioning

AEO / GEO thought leadership"GEO is not replacing SEO"
Citation Targets

Six assistants, one strategy.

The citation strategy and measurement spanned every major AI assistant a health question might reach.

ChatGPTOpenAI
Perplexityanswer engine
GeminiGoogle
Google AI OverviewsGoogle Search
ClaudeAnthropic
CopilotMicrosoft / Bing AI
By the Numbers

The infrastructure that shipped.

Infrastructure and process · verified from vault verified-metrics + Slack record. Quantified citation outcomes deliberately excluded.

AI platforms targeted
6
ChatGPT · Perplexity · Gemini · AI Overviews · Claude · Copilot
sameAs profiles
8
deployed site-wide · Organization + Person (commit 7559a87c23b0)
FAQPage coverage
98.7%
686 / 695 posts · the format AI parses for Q&A
llms.txt
LIVE
on production · published via Rank Math with entity context
Deployed
Person / Organization entity separation live in production (verified on Dev2)
Complete
medical-reviewer data audit — reviewer identity prepared for reviewedBy markup
In motion
movement toward MedicalWebPage-grade markup across the content library
Before / After

Three shifts toward being citable.

Entity graph
BeforeCombinedambiguous schema
AfterOrg + Personclean, separated
Machine legibility
BeforeNoneno llms.txt
AfterLiveentity file in production
FAQ schema coverage
BeforePartialgaps in Q&A markup
After98.7%686 / 695 valid
Results

What shipped — and what stays honest.

The proven claim is the infrastructure and the measurement approach. Quantified citation counts are held back until the methodology is separately reviewed.

● What was shipped · verified infrastructure

A health brand made materially more legible and trustworthy to AI search.

  • llms.txt entity file live in production via Rank Math.
  • A2A / NLWeb machine access and AI-crawler directives.
  • Clean Organization / Person entity graph with 8 sameAs profiles.
  • 98.7% FAQPage coverage; movement toward MedicalWebPage schema.
  • Medical-reviewer audit feeding author / reviewer authority signals.
  • Citation-tracking method spanning the six major assistants.
▲ Measured citation presence · methodology under review

Late-stage reporting did track AI-citation signals alongside SEO KPIs.

  • Reporting indicated the brand was surfacing in AI answers.
  • The definition of “citation” and the measurement methodology are not yet reviewed.

Quantified citation figures are deliberately withheld from headline use until the measurement definition and methodology are separately reviewed. The proven claim is the infrastructure and the measurement approach — not a specific citation count.

▲ Engagement context · 4-year · NOT this workstream alone

The broader Dr. Berg engagement (May 2022 – Apr 2026).

  • Organic traffic grew ~57K → ~181K monthly visits (+217%, Ahrefs full-period).
  • $8.86M in GA4 organic revenue attributed (Jul 2023 – Mar 2026).

Engagement-wide context, not a result of the AI-search build. The AEO/GEO work was one workstream inside the four-year engagement.

My Role

I set the direction and designed the stack.

I set the direction that AI search is a technical-SEO discipline for this brand, and built the infrastructure to match.

I designed what the llms.txt should say, which entity and authorship signals to deploy, how to separate the Organization and Person schema, how to prepare the medical-reviewer data for review markup, and how to measure citation presence across the six assistants. Code changes to the WordPress VIP repo were routed through the client’s dev team under a read-only-repo policy; I owned the strategy, the schema design, the measurement approach, and the review of what shipped.

Skills Demonstrated

What this took.

Answer-engine optimization (AEO)Generative engine optimization (GEO)AI-search infrastructure (llms.txt, A2A, NLWeb)Entity SEO & Knowledge Graph engineeringOrganization / Person separation, sameAsMedicalWebPage & FAQPage schema / JSON-LDE-E-A-T author & medical-reviewer markupAI-citation tracking & prompt auditingTechnical SEO on WordPress VIP / Rank MathAEO / GEO thought leadership

Want your brand answerable to AI search — before your competitors are?