A sampling system,not a universal AI rank.
AIBeam measures what supported provider APIs returned for a defined prompt set at a defined time. It preserves that evidence, separates interpretation from capture, and reports trends with the limits of stochastic AI answers made explicit.
Define the sampling frame
A saved prompt, prompt version, provider, project, and cadence define each observation. Keeping that frame explicit makes later comparisons inspectable.
Capture before interpretation
AIBeam calls the supported provider API, records attempt context, and preserves an accepted raw response envelope before analysis begins.
Analyze with a version
Brand occurrences, citations, competitor occurrences, and sentiment are tied to an analysis version so a method change does not silently rewrite history.
Aggregate traceable facts
Daily read models summarize accepted, analyzed responses while retaining links back to the response and analysis that produced each fact.
Primary visibility measure
Brand-present response rate
For the selected project, provider, and date window, visibility is the share of accepted responses in which the target brand is present under the selected analysis version.
The denominator matters. Failed attempts, exhausted observations, and responses outside the selected analysis are not quietly treated as brand-absent answers.
What this evidence cannot establish
These limits are part of the method, not fine print. Reports should be read as documented samples of provider behavior—not a census of every user, interface, or future answer.
Provider API responses can differ from the consumer chat application, including browsing, personalization, location, experiments, and interface-specific behavior.
Model output is stochastic. Repeating the same prompt can produce a different answer, brand set, wording, or source list.
A response-local occurrence position is context for that response, not a universal or deterministic AI ranking.
Citations are reported only when the accepted response includes native source metadata or extractable links.
AIBeam does not currently measure Google AI Overviews or AI Mode, and it does not infer them from another provider.
Monitoring observes change; it does not prove that AIBeam or a specific content action caused a visibility outcome.
Compare trends with the evidence attached.
Choose a prompt set, understand the credit budget, and keep the sampling frame consistent enough for changes to be interpretable.