How wrong is AI about you — and how long until it’s fixed?
Every AI engine holds a picture of your entity — some of it wrong, all of it aging. Entidex measures per-engine accuracyagainst your verified record, tracks how many days each engine is running behind the truth, and publishes the correction they'll adopt next. ChatGPT, Claude, Gemini, Perplexity and Grok.
AI Visibility is the supporting stat, not the story. The verified record, the AI engines, and the live X conversation rarely agree — we score the gap, name the wrong field, and close the loop.
Five engines. One entity. Do they agree?
Entidex reads ChatGPT, Claude, Gemini, Perplexity and Grokfor every entity, scores how each one describes you, and flags where they diverge — or repeat a fact that's out of date.
- ChatGPTOpenAIAlignedDescribes you correctly
- ClaudeAnthropicAlignedCites your own content
- GeminiGoogle DeepMindDivergesWrong category
- PerplexityPerplexity AIAlignedMatches the record
- GrokxAIStale factRepeats an outdated fact
One engine is repeating an outdated fact. Entidex measures the gap against your verified record and emits a signed correction the engines can read.
source signals resolved to exactly one canonical identity
An entity is any brand, person, organisation, or concept that can be uniquely resolved — with a canonical identity across all AI and data sources. Entidex resolves and monitors entities so AI platforms describe yours accurately.
Three pillars of entity intelligence
One canonical entity. Every public signal forming around it. Continuously observed, never manipulated.
Monitor
See what the world says
AI engines, social platforms, review networks, code repositories, news, search, geographic and self-declared signals — all watched continuously, all resolved to one canonical entity.
Understand
Decode the narrative and intent
Cross-source consensus, sentiment triangulation, narrative themes, citation tracking, and the gap between how an entity describes itself and how the world describes it.
Act
Move before perception hardens
Drift signals, momentum tracking, competitor proximity, authoring guidance grounded in observed evidence, and white-label reporting for agencies and analysts.
Intelligence Pipeline
Observe
Query AI platforms
Collect
42+ data sources
Analyse
Extract signals
Score
0–100 visibility
Interpret
Actionable insights
The truth gap
The distance between what's verifiably true about your entity and what AI engines actually say. Entidex fact-checks every asserted claim against sourced evidence — and shows you exactly where each engine drifts.
Lower is better. 3 of 5 sampled claims diverge on at least one engine — two are materially wrong.
The Entity Knowledge Statement
Every gap becomes a signed, machine-readable correction the AI engines can actually ingest — published as schema.org ClaimReview and offered through the MCP Correction Channel. Entidex is the canonical place to set the record straight: one protocol LLMs read and correct through.
// schema.org ClaimReview · Ed25519-signed { "@context": "https://schema.org", "@type": "ClaimReview", "claimReviewed": "Stripe is valued at $95B", "reviewRating": { "ratingValue": 2, "bestRating": 5, "alternateName": "Outdated — verified $65B (2024-Q2)" }, "citation": ["press:reuters/2024-03"], "author": { "name": "Entidex" } }
Each statement is Ed25519-signed and crawlable as Markdown, JSON-LD or JSON — ground truth an engine can trust and re-fetch.
LLMs read and submit corrections through one protocol (entity_submit_feedback). Reviewed in the correction inbox before it ever ships.
Every corrected claim cites the exact evidence and carries a 95% confidence interval. Replayable, auditable, never a black box.
An analyst, not a dashboard
Milo reads the evidence and tells you what it means — in plain language, first person, always cited. Three reads on every entity: the opening picture, the per-engine pattern, and the single highest-impact gap to fix.
Stripe reads strongly across all five engines — visibility 82, trust 88. But the picture isn’t clean: two engines are quoting a valuation a full year out of date.
Claude is your most accurate reader at 96%. Grok lags at 71% and is the one narrowing your category to “online payments”. This is an engine problem, not a perception problem.
Strongest lever: publish a signed Knowledge Statement with the current valuation and category. I’d expect ChatGPT and Grok to converge within two refresh cycles.
Every entity. Every surface. One read.
Six scores resolved from cross-surface observation — not surveys, not estimates.
AI Visibility
How prominently AI engines surface this entity versus its cohort — across ChatGPT (OpenAI), Gemini (Google DeepMind), Claude (Anthropic), Perplexity (Perplexity AI) and Grok (xAI).
View live dataSentiment
Cross-surface sentiment polarity merged from AI engine outputs, review networks, news, and community — with source agreement shown.
View live dataShare of Voice
Entity mention frequency as a proportion of all category-peer mentions across observed AI outputs. Normalised within vertical.
View live dataDecision Survival
How often the entity survives multi-turn AI conversations after budget objections, competitor comparisons, and negative framing.
View live dataCross-surface Consensus
Whether Trustpilot, Reddit, Google Maps, and AI engines are telling the same story. Low consensus is a signal in itself.
View live dataLifecycle Stage
One of nine states — from emerging to legacy — derived from visibility trajectory, drift score, and cohort-relative position.
View live dataMore signals most tools never show you
Visibility is the start, not the story. Every entity carries a deeper read — and it all composes over one canonical identity.
How much of what each AI engine says about you is actually right — per-engine accuracy against your verified record, down to the exact wrong field.
You fixed the record — how long until every AI engine catches up? Per-claim lag between verified reality and what engines still repeat.
Does your brand survive 3–5 turns of AI pressure — budget objections, competitor comparisons — or get dropped?
A nine-state classifier — catch entities in breakout, fragmentation, or decline before the market does.
One confidence-weighted credibility number across every surface. People and podcasts no longer score zero.
Real-time X social signal — a 24–72h lead over press, with an organic-vs-bot read.
Mistaken-identity detection — how often an AI confuses your person-entity with the wrong namesake.
Rank against a live cohort — Visibility, Sentiment and SoV deltas vs the cohort median.
A Claude-synthesised, intent-shaped brief on demand — every sentence cited to the run that produced it.
How each engine positions you — per-engine mention, recommendation and citation rates, plus cross-engine coherence.
The financial surface: live listing and quote data for public companies, folded onto the entity next to every other signal.
42+ live collectors. 8 signal categories. One canonical entity.
Every public signal forming around your entity — across the web, AI engines, communities and registries. Every collector below is wired up and running today. See the full collector list →
AI engines
How the five generative engines describe, recommend and cite your entity.
Identity & structural authority
Canonical identity and structural credibility across registries and knowledge bases.
Reputation & reviews
Customer-facing reputation networks where perception is recorded directly.
Community & social
Where people actually discuss you — developers, traders, fans, critics.
News
Press coverage and editorial framing across global news APIs.
Search & technical
Search visibility, AI Overviews and the technical health that backs both.
Geographic
Real-world location signals for places, venues and physical entities.
Sentiment NLP
Multiple independent NLP engines for triangulated sentiment.
Do your sources agree?
Your Trustpilot rating, your Google Maps reviews, the Reddit thread three pages deep, and what AI engines now repeat about you — these rarely tell the same story. Entidex merges them into a confidence-scored sentiment, surfaces where they disagree, and tells you which source moved first.
Trustpilot
Customer reviews
Google Maps Reviews
Local reputation
Reddit + Hacker News
Unfiltered community
Detect the gap before a customer or journalist does. How the cross-surface consensus index works →
Entity intelligence as an API
One request. All six canonical scores, confidence intervals, and raw evidence.
curl -s https://entidex.com/api/v1/entities/spacex/insights \
-H "Authorization: Bearer $ENTIDEX_API_KEY" | jq .scoresFour ways teams use Entidex today
Start with a free scan — then watch one entity from £9/mo.
Know what AI says about your business
Before your customers do. Snapshot in minutes, monitor in hours, fix what is wrong in days.
Explore Due diligencePre-deal entity intelligence
Investors, recruiters, and M&A teams: an entity intelligence view your IC, hiring panel, or board can defend.
Explore Rebrands & pivotsA rebrand isn’t complete until AI agrees
Track the transition chain across surfaces until every AI engine is citing the new identity.
Explore Social channelsWhich creators actually move the narrative
Authenticity scoring and cross-surface lift for creators, social accounts, and partner channels.
ExploreCore Principle
Entidex observes what AI says about you — it never manipulates it. Every signal is derived from observing AI-generated outputs; we never game perception, astroturf, or inject unverified content into any model. When the record is wrong, we publish signed, evidence-backed corrections with full provenance — corrections AI engines can independently verify and choose to adopt.
Every source. Every plan.
Simple, transparent pricing
No add-ons for ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google DeepMind), Perplexity (Perplexity AI), Grok (xAI), Reddit, or any collector. From a free scan to enterprise — pick a tier and scale up any time.
Monthly pricing shown — save 25% when billed annually. Need more than your plan includes? Add extra monitored entities at £9 each, any time. USD shown for reference — all charges are billed in GBP.
- Full evidence picture for one entity
- Consensus map across surfaces
- L1 + L2 signals
- 1 entity, weekly refresh
- Drift + alerts + X real-time
- MCP correction channel
- 3 entities (you + 2 competitors)
- X real-time + cross-surface drift
- AI Overview citation tracking
- Full API + MCP — every dataset channel
- 1,000 free credits/mo
- All six data products (L1–L4)
- 5 entities, 3 seats, daily refresh
- Decision Survival Score included
- Cross-source consensus + drift
- 25 entities, daily refresh, 5 seats
- White-label PDF exports + API write
- Cohort Comparison + priority support
See all tiers & full signal coverage
Free tier, Enterprise plans, exact credit allowances and direct Stripe checkout are on the full pricing page.
Drop Entidex into your AI client
Hosted endpoint. Wires Entidex into Claude Desktop, Cursor, and any MCP-aware client — authenticate with your API key.
npx -y mcp-remote https://entidex.com/api/mcpEvidence under every claim
Provenance on every claim
Every score links back to the probe run that produced it.
Append-only archive
Historical readings are immutable — replayable, never silently rewritten.
Confidence intervals on every score
Sample size and 95% CI shown, not hidden.
GDPR Article 17 honoured
Right-to-be-forgotten flow for person entities.
See what AI says about you
Run a free scan \u2014 no signup required. Resolve your entity and read its live AI Visibility, Sentiment and Share of Voice.
Run a free scan