About RefinedAI
RefinedAI measures how AI assistants describe your brand, and shows you the answer they actually gave so you can check the measurement yourself.
What the product does
Buyers increasingly ask an AI assistant before they open a search engine. RefinedAI treats those answers as a measurable surface. You give it a brand, its domain, and the countries you sell into. It builds a set of tracked phrases — the questions a buyer would plausibly ask — and sends them, on a schedule, to the major AI assistants.
For each answer it records who was named, where in the answer they appeared, how they were framed, and which sources the model cited. The result is a per-market picture of whether the AI names you, names a competitor, or names nobody at all.
Which models it asks
Tracked phrases are sent to assistants from OpenAI, Anthropic, Google (Gemini) and Perplexity. Google AI Overview and search results are retrieved through DataForSEO, and cited pages are fetched through Firecrawl when a run needs to verify them. Each model is queried directly with the market's country attached, because the same question answered from Singapore and from Australia does not return the same brands.
What it reports
- Mentions — whether the brand, and each tracked competitor, was named in the answer.
- Position — where in the answer the mention fell, since the first name in a list is not equivalent to the fifth.
- Sentiment and framing — how the model characterised the brand, not merely that it appeared.
- Citations — the URLs the model cited, attributed to the entity they supported, so you can see which pages are earning the answer.
Why we show the raw answer
Every measurement in RefinedAI is derived from a specific answer, and that answer is stored verbatim alongside the exact prompt that produced it. You can open any run and read the literal text the model returned.
This is deliberate. Scores that cannot be traced back to their evidence are easy to produce and impossible to trust. If our analysis reads a mention as positive and you disagree, the source text is one click away and you are free to conclude that we got it wrong.
What it is not
AI answers are generated by third-party models and vary between runs, between accounts, and over time. A RefinedAI run is a point-in-time sample of what a model said when asked a specific phrase from a specific country — it is not a ranking, not a guarantee, and not a measurement anyone can promise to move.
Who builds it
RefinedAI is a product of Aemorph. Aemorph is also the publisher of the LLM Visibility Index, and is excluded from that index's public rankings — see the methodology for how that exclusion is enforced.
[Aemorph to supply: company background — legal entity name, registered address, founding date, team size and any other corporate detail to appear here]
[Aemorph to supply: contact route for enquiries about this page (email address or contact form URL)]