Ask ChatGPT, Claude, Gemini, Perplexity, Grok about the same brand and you will get five overlapping but different answers. Cross-engine consensus measures how much they agree — and turns the places they split into a map of where your picture is weakest.
Every engine's account is resolved to one canonical entity, so you are comparing like with like.
The agreement between engines becomes a single Consensus / Divergence Score, tracked over time.
Divergence is attributed to the engine and the claim it splits on — not just a number, a target.
When ChatGPT, Claude, Gemini, Perplexity and Grok agree about your brand, they are almost always reading from the same consistent, authoritative sources. When they diverge, at least one of them is working from a different picture — stale training data, a thin retrieval, or a contradictory source it weighted too heavily. Low consensus is the single most reliable early warning that a confident wrong answer is coming.
Measuring it directly changes the work. Instead of probing one engine at a time and hoping to stumble on the discrepancy, you see the outlier engine and the exact claim it splits on — and you fix the source that is pulling it out of line. As the engines re-read on their next cycle, consensus rises and the gap closes.
Watch consensus move on the live demo · read the methodology · see the measured cohort answer: do AI engines agree with each other?
It is a measure of how much the AI engines agree about an entity. Entidex resolves every engine’s account of your brand to one canonical entity and scores the agreement between them as a Consensus / Divergence Score. High consensus means ChatGPT, Claude, Gemini, Perplexity and Grok are describing you the same way; high divergence means at least one of them is working from a different, usually weaker, picture.
Each engine is trained on a different mix of sources and retrieves differently at answer time, so the same entity lands in a slightly different position for each. They can disagree on founding date, category, ownership, even sentiment. The disagreement is not random — it points to where the underlying sources are thin, contradictory, or out of date.
Divergence is a signal, not a verdict. Low consensus tells you the engines are not reading from a consistent, authoritative picture — which is exactly when a confident wrong answer is most likely. By measuring divergence directly, you can see which engine is the outlier and which claim it is splitting on, instead of discovering the problem one prompt at a time.
Prompt monitoring tells you what one engine answered to one prompt on one day. Cross-engine consensus resolves all of those answers to one entity and measures the agreement structurally — across engines, across surfaces, and over time. The opinion is the symptom; the consensus is the diagnostic.
Yes. Run a free scan to see how each engine describes your entity and where they diverge, or watch consensus move across the tracked entities on the live demo.