The measurement layer

See exactly where AI recommends you, and where it does not

You cannot fix what you cannot see. The scan is the measurement layer underneath everything else TofuBofu does. It asks all 6 major AI engines the questions your buyers actually ask, records who gets named, and turns the result into a score you can track month over month.

Why a scan, and not just a Google search

When you type your category into Google you get ten blue links and you can see where you rank. AI engines do not work that way. They synthesize a single answer, name two or three companies, and move on. There is no page two, no visible ranking, and the answer changes with the exact phrasing of the question. The only way to know whether you are in that answer is to ask the question the way a buyer would, on each engine, and read what comes back.

That is what the scan automates. Instead of you manually opening six chat windows and typing variations of the same question, it runs a structured set of buying-intent queries across all 6 AI engines (ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, and Bing Copilot), then parses every answer to check whether your brand was actually named. Not whether it could have been. Whether it was.

How the engines decide who to name

Modern AI answers are produced by retrieval-augmented generation. The model does not recite your homepage from memory. It retrieves relevant documents from an external index, then generates an answer grounded in what it retrieved. The original research describes this as combining a model's internal knowledge with a separate, updatable store of documents so the output is more specific and more factual.

The practical consequence for you is simple. To be named, your information has to be retrievable and unambiguous at the moment the engine builds its answer. Content that is structured, specific, and corroborated across several sources gets retrieved and cited. Generic marketing copy does not. The scan shows you the downstream result of that mechanism: named, or not named, for each query.

A score built from real buying questions, not a vanity number

Not every query matters equally. A buyer asking which vendor to hire is far closer to a purchase than someone asking what a category means, so the scan focuses on bottom-of-funnel buying questions, the ones closest to a purchase decision, rather than diluting the score with generic awareness queries. A firm that talks a big game in its own content but is absent when AI is asked who to hire will see that gap in the score immediately.

Every result is verified server-side. The engine's answer text is checked for your actual brand, including domain-style and spacing variants, so a passing mention counts and a near-miss does not inflate the number. The output is a score you can defend and a ranked list of the specific queries where you are losing, which is exactly what the content solutions are built to close.

What you get