Keep your brand ahead inThe AI era
From the infrastructure AI needs to the content AI reads.
Answer engines we monitor and agents that can connect
- ChatGPT
- Perplexity
- Google AI
- Cursor
- Claude Code
- VS Code
- Windsurf
- Copilot
The core
Measure. Fix. Verify.
Your buyers have stopped scrolling ten blue links. They ask ChatGPT, Perplexity, Google AI — and act on the answer they are handed. If those engines can't read your site, or don't name you as the source, you are not slipping down a ranking. You are absent from the conversation.
And the rules keep moving — new bots, new protocols, new models, no changelog. We measure both halves, whether AI can use your site and whether engines cite it, then re-measure as the ground shifts. You keep building. We keep watch.
Connect your site to AEOForged
See who can ship- WordPress
- Shopify
- GitHub
- Webflow
- Wix
- HubSpot CMS
- and more
Code in a repo? Install the AEOForged GitHub App. We open a pull request. We never force-push your default branch. Squarespace and Framer get a publish brief — they have no write API we can use.
Install the GitHub AppAI readiness / technical foundation
Can AI systems actually read your site? Robots access, crawlability, structured data, extractable answers on the pages that matter, identity signals so engines know who you are. If the site is hard for us (or them) to read, nothing else sticks.
Authority & disambiguation
Engines need a clear picture of your brand — not whatever else "brand" already means on the web. That is entity work: about pages, verified profiles, consistent facts, off-site corroboration over time.
Monitoring
We pick a board of real questions (branded + the buyer questions you care about) and re-check them in live engines. A slot turns green only when a page of yours is actually cited. "Not cited" is a result we report, not something we paper over. Branded recognition and category citations are measured separately — they are different problems.
Content & distribution
When monitoring shows a gap, we research that question and your agent writes the page against the evidence — plus kits for LinkedIn / X / newsletter so the work can travel. We package; you post. We don't take social logins and we don't promise reach.
LAYER 01 · AI READINESS
Can AI systems actually read — and operate — your site?
Layer one is the technical foundation. We probe the live site for crawler access, extractable answers, structured data, and agent operability. If machines can't reach or parse you, citations and authority work don't stick. Your agent (or team) applies the fixes; we measure with verify-against-baseline — nothing is “done” until the re-check says so.
Where you stand
Free audit · fixes locked until you pick a package
AEO / content
77
Grade B · editorial
AI-ready foundation
75
15 of 20 checks
On-site authority
92
Entity · topical checks
Illustrative · sampling disclosed on every real report
Agent operability probe
Live fetch with real AI-agent user-agents
5/5 agent probes reached homepage
6/7 forms appear agent-operable
N/A — no booking flow detected
Commerce schema signals present
Observed operability today — not agent purchase volume or projected revenue.
Capability scorecard
Deterministic — measured, not modelled
- AI crawler accessGood
- Direct-answer leadsGood
- Structure & formattingWeak
- Definition blocksWeak
- llms.txtGood
- Booking schemaN/A
Protocol signals
28 live · 4 N/A
Discovery
8/8
Covered
Content
9/9
Covered
Auth
3
2 N/A
Protocols
4
2 N/A
Free audit standing
Three measured standings from the live site: content AEO, AI-ready foundation checklist, and on-site authority. Itemized fixes stay locked until you choose how to apply them.
Capability scorecard
Deterministic checks — good, weak, or missing — across access, machine meaning, answer-ready content, and trust. No model invents the ratings.
Agent operability
We fetch with real AI-agent user-agents and score what agents can do today: read, forms, booking, transact. Capabilities that don't apply to your site type are N/A — not failures.
Protocol signals
Discovery manifests, content structure, auth discovery, and agent protocols — live, auto-fixable, manual, or not applicable. Compatibility remaps (Cloudflare / Lighthouse / FORKOFF style) are our signal map, not third-party live scores.
LAYER 02 · AUTHORITY
Engines need a clear picture of your brand — not a lookalike.
Layer two is entity work: about pages, verified profiles, consistent facts, and off-site corroboration measured over time. Your agent applies the identity fixes; we re-probe the chain and recognition. Authority compounds slowly — we report the binding constraint, not a fake DA lift.
Authority profile
Seven measured signals · how engines resolve your brand
Where the next effort goes
Entity identity is the current bottleneck — signals downstream of it cannot convert until it clears.
- Entity identityMeasured54
- People & expertiseMeasured60
- Topical depthMeasured100
- Primary evidenceMeasured40
- DistributionMeasured100
- Off-site corroborationPartial0
- AI recognitionNo data yet—
Authority programme
Measured phases · P3 plan-only
93%
Measured progress
P0: Entity Foundation
79% measured
P1: Topical Authority
100% measured
P2: Distribution
100% measured
P3: External
Plan-based (off-site)
Entity recognition
Does AI know who you are?
- PerplexitySelf-citedRecognized
- ChatGPTSelf-citedRecognized
- GeminiSelf-citedRecognized
100%
Recognition rate
100%
Self-citation rate
Measured standing in live engines — not a citation guarantee for category queries.
Entity chain score
2/7 links verified
36 / 100
sameAs only counts when URLs resolve live · Wikidata is draft until measured
Authority profile
Seven measured categories — entity identity through AI recognition — plus an archetype and a binding constraint that names where the next effort goes. Scores are derived in code, not invented by a model.
Programme phases P0–P2
Entity foundation, topical authority, and distribution are live checklists from your site. External validation (P3) is plan-only until placements are verified live — we never pretend off-site work is measured when it isn't.
Entity chain & recognition
Wikidata, sameAs, socials, mentions, knowledge panel, content authority, AI recognition — each link verified or flagged. Recognition probes real engines for your brand; competitor compare never invents their recognition.
Off-site receipts
Verified-live placements, linked and unlinked mentions, social consistency — sample-scoped, re-checked by the system. Your agent wires the identity; we measure what is actually live.
LAYER 03 · MONITORING
A board of real questions. Green only when a page is cited.
Layer three is measurement. We re-check your branded and goal questions in live engines, report every miss, and never call a brand mention a citation. Branded and category rates stay separate — they are different problems.
Your citation board
The questions you chose, put to real AI engines on a schedule. Green means a page citation — never an estimate, never a mention folded in.
22/58
Questions cited
Branded lane
19/24
50 slots
Goal lane
3/34
50 slots
Citation rate
15%
roughly 11–19%
Last check
7 Aug
Live engines
Board rows
Branded + goal · page-tier green only
22 cited · 36 open
- Cited — your page
AEOForged before and after case study
Page cited by Perplexity, ChatGPT
- Cited — your page
AI ready content
Page cited by Perplexity
- Not present yet
AEO agency for B2B brands
No page citation yet
- Not present yet
best tools to track ChatGPT citations
No page citation yet
Milestones
Measured the day engines make it true — not the day work ships
AI cites your pages when asked about you by name.
19 of 24 branded questions cited.
You own your brand questions — every branded slot cites a page.
19 of 24 branded questions cited.
First unbranded citation: AI recommended you on merit.
3 of 34 goal questions cited.
AI cites you on half the unbranded questions we track.
3 of 34 goal questions cited (half is 17).
Citation rate by lane
Never blended — branded and goal stay separate sentences
Branded questions
37%
Cited 34 of 92 checks
Goal questions
3%
Cited 5 of 176 checks
By AI engine
Perplexity
29%
Gemini
29%
ChatGPT
13%
Citation board
A board of real questions — branded plus the buyer questions you care about — re-checked in live engines. A slot turns green only when a page of yours is cited. Mentions and “not cited” stay visible.
Two lanes, two problems
Branded recognition and category (goal) citations are measured separately. Winning your own name proves the plumbing; goal-lane green is the authority signal the programme exists to move.
Milestones & answer coverage
Milestones are measured statements about the board — reached the day engines make them true. Answer coverage joins each goal question to whether a page holds an extractable answer — deterministic, not a model guess.
Rates, engines, sector
Citation rates by lane and by engine, optional sector benchmark on the same sweep questions, and share of voice on goal questions only — brand-name prompts never inflate SOV.
We run the same board on ourselves. Live AI citation rates for aeoforged.com — branded vs generic, updated as daily checks land, limits stated.
LAYER 04 · CONTENT & DISTRIBUTION
Research the web. Ground the draft. Score what you write.
When monitoring shows a gap, content work starts with our web research tool — live search, structured synthesis, citations limited to URLs we actually retrieved. Save runs in a research library, reuse or merge them for richer briefs. We research and score; your agent writes.
Queries used
Entities · 29
Questions · 11
What scoring methodology separates measured AEO from rebadged SEO tools?
Claims with sources · 13
- moderate
Multi-LLM tracking measures citation breadth across engines, not a single proxy.
nogood.io/…
- anecdotal
Answer-shaped sections with a direct lede improve extractability in AI answers.
reddit.com/r/…
Illustrative product UI — same shape as the research panel (claims + source URLs).
Grounded web research
The research tool searches the live web, then synthesizes entities, claims, data points, open questions, competitor angles, and risk flags. Every claim is tied to a retrieved URL — invented footnotes get stripped.
Save, reuse, and merge
Runs land in a workspace research library. Pin research to an article, reuse a prior pack, or merge several runs when you want a deeper brief — compounding is deliberate, not automatic magic.
Answer-shaped outline
Sections carry intent, a direct-answer lede, and E-E-A-T slots filled from that research pack. Sources travel with the outline into the draft.
Write, then score
Your agent usually writes (or use hosted draft tools). Eight AEO dimensions (0–100) flag weak structure, extractability, and entities — feedback for the draft, not a citation guarantee.
Outline → draft → eight-dimension score
Question-shaped H2s and direct answers come from the research pack. After the draft you see structure, entities, extractability, and more — then target weak dimensions before publish. Scores are publish-readiness feedback, not a promise that engines will cite you.
- Competitor angles and risk flags surface gaps before you write
- Agent brief / export when writing in Cursor or Claude Code
- Grounding gate: inline citations must match retrieved research URLs
Heuristic AEO checks — feedback for the draft, not a citation guarantee.
Then package the distribution kit
After a live URL you're happy with, we turn the article into channel-ready copy — LinkedIn post and article, X thread and X Article, carousel slides, YouTube script, and a newsletter section. We package; you post. We never take social logins, and we never claim reach we don't control.
- X
- YouTube
- Newsletter
Ordered deployment plan: canonical → socials → native longform with link-back to your domain.
Why our numbers can be trusted
Every claim we make is falsifiable.
Probes hit live answer engines; verification re-fetches your page and re-scores it against a pinned baseline — so we cannot grade our own homework. Read the methodology.
The system grants “done” — not us
Nothing is verified because a person or a model said so. A re-probe fetches the live page, re-scores it against a pinned baseline, and is the only thing that can mark work verified.
Before/after is a diff, not a memory
Every audit persists an immutable snapshot. The lift we report is a comparison of two frozen records — never a recollection of what we changed.
We record the misses
Every check where you were NOT cited is stored alongside the wins. If a vendor's only data is wins, that's marketing, not measurement.
Run-to-run variance, published
Identical prompts, identical engines, same day — we measure how often answers flip, and we never call movement inside the noise floor a win.
Key takeaways
AEOForged is the measured AEO layer that checks real engines and verifies fixes — not a ranking promise.
- 01
Answer Engine Optimization is structuring content so AI answer engines cite your brand as a source.
- 02
AEOForged measures citations on ChatGPT, Perplexity, and Google AI and stores the misses alongside the wins.
- 03
Verify-page re-fetches a live URL and re-scores it against a pinned audit baseline — the system grants verified, not a human checkbox.
- 04
The citation ladder has three rungs: none, domain (brand-name citations), and page (category-query citations gated by entity authority).
- 05
The self-audit proof dossier (measured 2026-07-07) recorded 0 of 12 assets cited and an entity-chain score of 29/100 — thin sample, negatives kept.
Frequently asked questions
What is AEOForged?
AEOForged is a measured Answer Engine Optimization service for brands and agencies.
It audits and tracks whether ChatGPT, Perplexity, and Google AI cite you, records misses as carefully as wins, and verifies fixes against a pinned baseline. Packages run from a free diagnostic through Fix Programme and monthly Monitor / Dominate retainers.
What is Answer Engine Optimization?
Answer Engine Optimization (AEO) is structuring content so AI answer engines cite your brand as a source, not rank a page in a list.
AEOForged measures those citations on ChatGPT, Perplexity, and Google AI — and records the misses — instead of promising SEO-style rankings.
How does AEOForged measure citations?
AEOForged measures citations with live checks against real answer engines for tracked queries, storing both cited and not-cited outcomes.
Fixed pages are re-scored against a pinned audit baseline. Misses are stored alongside wins; movement inside measured run-to-run noise is never called a win.
Our honesty bar
We sell measured readiness, delivered assets, and verified score lift. Nobody can honestly guarantee AI citations — including us. Numbers labelled “proxy” or “directional” are exactly that.