Skip to content
AVE StudioAI Visibility Engineering
PROPRIETARY STUDIO INSTRUMENT

KG Doctor · Examining the company data machines can reach

KG Doctor· GEO + KEOdetailed findings instead of a single score

Generative Engine Optimization

Being found, included and cited when an AI generates an answer — at answer time, from the live web.

Read the definition
+
Knowledge Engine Optimization

The verified, canonical, readable, citable record AI models learn from — so each new model is trained to know and recall your business correctly.

Read the definition
— the studio's own diagnostic instrument

Different data sources state contradictory things about you, so the AI can't be sure who you are. The KG Doctor uncovers where — and why — machine identification breaks down.

The
Machine equity

Machine equity is the business asset that arises from AI systems being able to recognize, understand, recall, recommend — and, on a user's behalf, transact with — a brand. Composed of three layers: rented reach (retrieval visibility), owned memory (trained model recall), and machine buyability (agent-side transactability). Measured honestly only as a portfolio of ranged, per-layer metrics — never as a single blended score.

Read the definition
loop

  1. 1 · MEASURE — where you stand and which questions matter (Citation Tracker, Selector)
  2. 2 · FIX — why the AI skips you, fixed in code (Gap Analysis, Gap Closure)
  3. 3 · ANCHOR — an unambiguous entity, present where models learn (Entity Anchor, Corpus Campaign)
  4. 4 · PROVE ↺ — closed-book baseline and the delta at every new model generation (Model Memory Audit, Release Audit)
→ AI TRANSACTION — when the agent buys (agentic commerce · roadmap)

KG Doctor and the Claim Coherence Engine — the studio's own instruments — serve the entire loop.

AVE Studio doesn't run short-term campaigns: we build our partners' machine equity, across model generations.

The problem

AI models identify your company from several structured data sources — Wikidata, the structured data embedded in your website, and company profiles, among others. If these state different founding years, addresses or other facts, or the links between them are missing, the machine cannot form an unambiguous picture of you. The KG Doctor works in three steps: it uncovers the symptom, produces the findings, then delivers a treatment plan.

KG Doctor works with structured facts — what machine sources (Wikidata, embedded structured data, company profiles) state about you. The prose marketing claims on your website are examined by its sibling instrument, the Claim Coherence Engine; both feed the same approved fact base.

The examination — in three steps

01The approved-facts registry

From machine-readable sources — your website, Wikidata and company profiles — we collect the claims made about you, then show the contradictions. You decide which value is correct; the decision is logged and becomes the official reference for every later check.

02Identifiability examination — 8 checks

We examine whether machines identify you unambiguously. We check the completeness of the Wikidata item, the quality of its references, the freshness of the data, and we compare the approved facts, Wikidata and your website's structured data. The goal: establishing whether all three sources state the same thing.

03What do AI crawlers reach?

We check which AI crawlers can access the site — GPTBot, ClaudeBot or PerplexityBot, for example. We verify whether the important claims appear in the page source without JavaScript, whether browsers and AI crawlers are served different content, and whether an llms.txt file exists and is adequate.

Behind the service: purpose-built software

Every service runs on an application we built in-house for exactly this job — not manual spreadsheets, not general-purpose tools.

Our diagnostic engine: an approved fact registry, eight identifiability checks, and a dual browser-vs-AI-crawler fetch comparison.

The findings — and why there is no score
What we deliver

findings, severity, verifiable evidence

Every finding carries a severity classification, from plain information to a work-blocking error. We attach the verifiable evidence: the affected URL, a machine snapshot of the page recorded at examination time, and a table of the discrepancies. A machine-generated suggestion can enter the final findings only after human review.

What we deliberately don't

a merged "knowledge-graph score"

A single score would hide what exactly needs fixing. A 61 doesn't tell you whether the Wikidata item is stale or the AI crawler can't reach your price list. Detailed findings, by contrast, translate directly into action.

The treatment plan

You get a step-by-step task list, in the required execution order. Every item states what to do, where to change it, and the expected effort. The Hungarian and English report is shareable. Our quality rule: no final client report ships while a severe open error remains that blocks closing the work.

Why it's different
01

It examines what the AI crawler actually reaches

We download the page raw, without JavaScript, then compare what a regular browser receives with what an AI crawler receives.

02

It measures against your own approved facts

No generic checklist: we examine whether your approved claims appear consistently on the surfaces machines read.

03

Diverging data is itself a finding

If the approved-facts registry, Wikidata and your website state different values, we present it as an itemized finding, with evidence.

04

Built on evidence, not opinion

Behind every finding stands a verifiable URL and the technical evidence recorded at examination time, so every single item in the report can be re-checked.

Where this instrument works

This instrument prepares the fact base for these paid modules:

A proprietary internal diagnostic instrument of the studio that powers the paid modules. Not a standalone, purchasable product.

Honesty note

The KG Doctor currently examines the approved-facts registry, identifiability, and AI-crawler access. Press and corpus analysis, and the model-memory panel, are planned features; the model's own knowledge is currently measured by the standalone Model Memory Audit. The fix is always human work: the KG Doctor delivers a diagnosis and an execution plan, but never edits Wikidata automatically. ---

Curious what AI says about you?

Start with the free entry check — no commitment.

Request your free check