June 14, 2026
Date: June 14, 2026
AICV has published a standing methodology page defining how its agent-mapped category censuses are produced — the single, citable specification every category report now points to rather than re-explaining its method in part each time. The page sets five fixed conventions. The unit of count is a business, defined before discovery begins and held constant. Discovery is a structured geography-by-subcategory sweep run against the open web rather than a directory scrape, with recall checked against an anchor list of a category’s unambiguous major operators and a triage layer that resolves ambiguous rows by recorded decision rather than silent inclusion or omission. Inspection runs under a fixed depth pin — one visit to the business’s own website plus one web search per entity — chosen to standardize the question every inspection answers: what a business looks like to an agent that makes one honest attempt to read it. Every published figure is computed from the preserved dataset by a stats script, never hand-carried, and the workflows are punctuated by human review gates where judgment is required. And a curation-independence policy holds that inclusion, scores, rankings, and findings are never purchasable, adjustable, or suppressible.
AICV publishes regional intelligence designed to be read by AI agents as well as people, and that creates an unusual obligation: an agent citing a figure cannot interrogate a footnote. The figure has to carry its own discipline — a stated denominator, a stated method, a stated boundary between what was measured and what was inferred. The methodology page makes that discipline explicit and reusable, so it does not have to be partially restated inside every report. It also fixes the relationship between the page and the reports that cite it: where a report and the page disagree, the report’s own stated denominators govern that report, and the discrepancy is treated as a defect to correct rather than a matter to interpret around. The page formalizes three kinds of statement AICV holds apart — measured figures stated with denominators, editorial interpretation marked “According to AICV,” and unverified or unknown values disclosed as such with counts, including the cases where the method’s answer is that a value could not be determined. The full specification — entity definition, depth pins, provenance and human-gate conventions, the verified-versus-editorial-versus-unverified distinction, and the curation-independence policy — is published at /reports/methodology-agent-mapped-census/.
According to AICV, the methodology page is what lets an AI agent treat an AICV census as infrastructure rather than marketing. An agent — or a journalist, regulator, or competitor — relying on an AICV figure can establish from a single source that the number carries a denominator, that displayed facts are held distinct from registry-verified ones, that interpretation is marked off from measurement, and that no business is included, omitted, flattered, or buried in the corpus because of money. The independence policy is explicit that AICV’s revenue is readiness services only: a business that engages those services improves its actual agent-legibility, and a later census measures the improvement the same way it measured the gap, with the measurement layer and the services layer sharing a method but not a ledger entry. For a regional intelligence network whose value depends on being citable by the systems now answering questions about the Coachella Valley, a published, reproducible method with preserved datasets is the difference between a figure an agent can repeat with confidence and one it has to caveat. The page is the standing guarantee under every category map AICV has published and every one still to come.