AThe AI Visibility Index

Developers

Cite the index programmatically

Every reading The AI Visibility Index publishes on a scorecard is also available as JSON — the same data, machine-readable, so agents, researchers and journalists can cite the index without scraping it. No key, no signup. Free to use with attribution and a link to the scorecard.

The scorecard endpoint

One GET per company, addressed by the slug in its scorecard URL:

GET https://www.theaivisibilityindex.ai/api/v1/companies/city-cigars

The response carries the headline AVI Score (60% Outcome, 40% Readiness), the Readiness percent from the latest crawl, and the Outcome reading — how often AI assistants named and recommended the brand across sampled buying prompts, with Wilson confidence intervals, the engines the reading actually ran on, and the corpus and judge versions that produced it.

What the API will never tell you

The same honesty rules as the human pages, enforced in the payload: a half that has not been measured is null, not zero — and a simulated run is never served as a measurement. When a simulation exists but is withheld, simulatedOutcomeExcludedis true so you can tell “not measured yet” from “we refused to publish a simulation”. The engines listed are the ones that reading ran on, never the index's current capability list. The method behind every number is public: how the AVI Score works.

Machine-readable interfaces

Attribution

Cite as “The AI Visibility Index” with a link to the company's scorecard. Every payload carries an attribution field and the canonical links. If you build something on this data, we'd genuinely like to hear about it.

Fair use

Responses are edge-cached for an hour — readings change at most daily, so polling harder than that fetches nothing new. There is no write access; the index is measured, not edited.