The honest instrument for AI search.
Most tools in this category sell certainty they do not have: scores to two decimals from a single sample, wobble reported as progress, numbers only the engines could possibly know. ClarAI is built the other way. We would rather tell you the honest, smaller truth than the confident, wrong one. Here is what we commit to, and how each commitment is wired into the product.
The category's loudest problem is false precision.
AI answers are not deterministic. Ask the same question twice and the wording, the order, and the brands named can shift. A tool that reports a single crisp figure from one sample, and treats every small movement as a trend, is selling a feeling of accuracy rather than the thing itself.
Statistical honesty is our whole wedge. It is not a tagline we add to the marketing; it is a discipline the product enforces on itself. The five commitments below are the shape of it.
- 01
We show the sample size on every score.
Every visibility number carries the count behind it: how many prompts we measured this run, across how many engines. A 100 percent that came from four prompts reads as four prompts, not as the whole internet.
What this means for you
You can defend the number in a meeting, because you can see exactly what it is made of.
- 02
We never claim access to an engine's internal metrics.
No third-party tool can see inside ChatGPT, Gemini, or Google's ranking systems, and we do not pretend to. We measure what is observable: the answers the engines actually give and the pages they actually cite.
What this means for you
Nothing on your dashboard is a black-box guess dressed up as a proprietary signal.
- 03
When a change is inside the noise, we say so.
AI answers vary from run to run. We track the run-to-run spread of your score, and when a move is smaller than that spread we call it steady, not a win and not a loss. A tie is a tie.
What this means for you
You act on real movement, and you do not lose a week chasing a number that only wobbled.
- 04
We measure through official engine APIs, and we say what that covers.
We sample three engines, ChatGPT, Gemini and Claude, through their official APIs and web-grounded search. Google AI Overviews is built but not switched on yet. That is a fair, repeatable read of what the engines answer. It is not the same as one logged-in person's personalised session, and we tell you that instead of implying we watch every user's screen.
What this means for you
You know precisely what the score represents, so you never mistake a sample for a census.
- 05
Simple view hides screens, never data.
The Simple view shows fewer surfaces. Free accounts stay in Simple; Pro and Agency accounts, and every account during its 7-day trial, can switch to Standard or Advanced at any time. Every view shows the same sample sizes, the same raw answers, the same honest numbers. Simple is fewer rooms of the same house, never a stripped-down copy of your data.
What this means for you
Starting simple costs you nothing in truth: the view changes which screens you see, never the numbers on them.
See the honest version of your own numbers.
Grade your site free and get a letter grade with the checks behind it, or create an account and watch what three AI engines actually say about your brand, sample sizes included.