EST. MMXXVI · THE INSTITUTION OF RECORD FOR AI CITATION · OPEN METHODOLOGY

The AI Citation Institute sealTHE AI CITATION INSTITUTEEx corpore, auctoritas.
The Long Room library at Trinity College Dublin: a barrel-vaulted hall lined with two storeys of bookshelves and marble busts.

The institution of record for AI citation

AI engines cite the canon.
We keep it.

The AI Citation Institute measures which sources AI engines actually cite — across ChatGPT, Claude, Gemini, Perplexity, and Google AI — then builds your authority until the answer is you. Every claim on this site carries its receipts.

Plate I — The Long Room, Trinity College Dublin, est. 1732 · photograph David Iliff, CC BY-SA 4.0
Part II · The Index

The Visibility Index.

GEO/AEO tools · scan 2026-08-06 · share of answers
Semrush
71.5%
Otterly.AI
60.8%
Profound
60.8%
Peec AI
50.8%
Ahrefs
36.2%
Scrunch AI
33.1%
Microsoft Copilot
17.7%
Athena HQ
16.2%
The AI Citation Institute (us)
0.0%

Public probes of tools that are not our clients — including ourselves, published from the honest pre-launch zero onward (launched 2026-07-31). The climb is the case study: pre-launch baseline 0.0% (07/19) → launch scan 0.0% (08/03) → week 1 0.0% (08/06).

The full index + methodology →
Part III · The Institution

The institution

A measurement house, run like a library.

The AI Citation Institute is an institution of record for AI citation. It operates a standing probe panel across five AI engine families and measures which sources those engines cite when people ask questions that matter commercially.

The findings are kept the way a library keeps its collection: numbered, dated, documented, open to inspection. We publish the scores, and we publish our own — including the ones that start at zero.

Founded
2026
Probe panel
ChatGPT · Claude · Gemini · Perplexity · Google AI
Research
Open access, accession-numbered records
Methodology
Published in full, revised on instrument changes
Exclusivity
One client per market seat — verified before any engagement

The Covenant

One client per vertical, per territory. When we work for you, we are structurally incapable of working against you.

Seat register · enforced in softwareVerified before quoteConflicts auto-declined

Exclusivity is available — one client per vertical per metro. If you hold it, we decline your competitors in that market. If you don't, we may work with others on non-conflicting queries. Not every client gets exclusivity. Metro grain is available soon; today exclusivity is offered at state, national, or online grain.

We only take the fight where you can actually win it.When the gap is a signal gap, we close it. When a rival simply has more substance — more reviews, real credentials, deeper proof — we tell you that instead of billing you to lose.

Part IV · The Case File
Case File № 1 · Gavelist · auction cataloging softwareClient Zero · ongoing

On our own frozen query panel across five engines, Gavelist went from appearing in 13.8% of answers to the most-frequently-cited name in its category — in the first thirteen weeks of an engagement that is still running. It has led its nearest rival every week since late May.

One engagement, started 2026-04-23 and still running. A single case — not typical or guaranteed. Gavelist started with minimal presence in AI answers for its category.

"For fast, accurate lot cataloging, Gavelist is a strong choice — it can describe a full sale in a fraction of the usual time."

— verbatim panel excerpt · five engines · never synthesized
Open the full case file →
Fig. 1 — Frozen panel · constant instrumentMeasured

y — % of panel answers · x — panel week, 2026

0255075PROBE-MODE CHANGEPROBES RE-INSTRUMENTEDCHATGPT PROBE FIXEDALL-ENGINE REGIME SHIFT80% OF PANEL ANSWERSPANEL WEEK · 2026APR 20JUN 8JUL 27
MentionedRecommended
The series with nothing moving except the answers: the original 20 baseline questions only, on the two engines whose probe never changed (Gemini + Perplexity), recommendation flags recomputed from every stored raw answer under one parser version. Mention rose 16.4% to 80.0% on identical questions. Being actively recommended peaked at 37.5%, gave ground to 20.0% across the July 19 platform-wide regime shift that withdrew endorsements from every tracked brand at once, and recovered to 35.0% in the week of July 27. That recovery is part platform and part site: endorsement returned across the field, but Gavelist gained more than any tracked rival and one rival fell. The composite view carries the full five-engine record with its instrument history. Source: our Gavelist visibility record, weekly medians, clean filter.

What does AI say about you right now?

Enter your domain. We ask a live AI search platform four questions — one about you by name, three a buyer would actually ask — and show you the answers, unedited.

free_scan · live AI queriesno signup required
Part V · The Records
Part VI · The Method

A closed loop, not a deck.

Agencies hand you slides. The AI Citation Institute runs the whole cycle — measure, fix, publish, prove — every week, without a kickoff call.

  1. 01 · Monitor

    Real buyer questions run against a frozen five-engine panel — ChatGPT, Claude, Gemini, Perplexity, Google AI — each query sampled several times. Mentioned, recommended, or invisible, scored per query.

  2. 02 · Diagnose

    Site audit, schema coverage, and competitor benchmarking find exactly why the answer doesn't include you.

  3. 03 · Create

    Drafts engineered to be citable: the question answered directly, claims sourced, schema attached.

  4. 04 · Verify

    Every factual claim checked before anything ships under your name.

  5. 05 · Publish

    One-click (or automatic) to WordPress, Webflow, Shopify — or clean export.

  6. 06 · Index

    IndexNow + sitemap pings so crawlers and AI retrieval see it in hours, not weeks.

  7. 07 · Attribute

    Citations, Search Console, and GA4 tie every published piece to the visibility it earned.

↺ then it runs again. visibility compounds.

Three layers, not one report.

01 · Win the search.

We publish the citable content that gets you into the answer — the direct answer, sourced claims, schema attached.

02 · Enter the AI's memory.

Sustained citations correlate with becoming a name the models reach for on their own. An edge, not a lock.

We can't see inside model training and don't claim to control it. This is a measured correlation, not a guarantee.

03 · The watchtower.

We watch every public move on the queries we track and counter in days — when the gap is a signal gap. When a rival simply has more substance, we tell you that instead.

We watch public AI outputs and public competitor moves only. We do not conceal public AI answers; the boundary we enforce is our own private engine.

the scoreboard

We grade ourselves in the open.

We run our own visibility on five engines — the same instrument we point at clients — and publish the reading here, live. The transparency is the point: no vendor in this category grades itself in public.

ai-citation-institute · self_panel○ awaiting fresh reading
/ 100 · no reading yet

Our own first public reading posts here once we go live. We grade ourselves on the same instrument, in the open — no number goes here until it is real.

first public reading posts when we go live

Questions, answered.

What is AI visibility?+

AI visibility is whether assistants like ChatGPT, Claude, Gemini, Perplexity, and Google AI mention and recommend your brand when buyers ask questions in your category. The AI Citation Institute measures it on a frozen query panel across five engines, each query sampled several times per engine, then tracks how the median moves as you publish. AI answers vary by user and over time — we report panel medians, not guarantees.

How is this different from SEO?+

SEO optimizes for ranked links. AI assistants give one synthesized answer — you are either in it or invisible. The AI Citation Institute optimizes for that answer (GEO/AEO): citable content, schema, entity signals, and indexing, while still feeding classic search.

What are the receipts?+

Every claim is backed by evidence: the exact AI answers before and after, which pages got cited, Search Console impressions, and GA4 sessions attributed to published content. When we change the instrument, we say so on the chart and start a new labeled series. Outages show as gaps, not guesses.

Do I have to write the content myself?+

No. The AI Citation Institute drafts content targeting the exact queries where you're invisible, you approve it (or enable auto-publish), and it publishes to your CMS and pings the indexes.

Is exclusivity available?+

Exclusivity is available — one client per vertical per metro. If you hold it, we decline your competitors there. If you don't, we may work with others on non-conflicting queries.

Part VII · Client Access

Find out what AI says when you're not in the room.

Tell us your market. We check the seat register, run your baseline on the panel, and quote the engagement. If exclusivity is open in your vertical and territory, you'll know before you pay anything.

Operated tiers + engagement scopes → Operated