Can you show me the exact AI-facing response for real URLs?

Yes. Every engagement includes a live, per-URL demonstration of the machine-layer response side by side with the origin page. A public example of this discipline: the hash-anchored baseline tables on the V Digital Services proof page, where every raw engine answer carries a SHA-256 of its verbatim text.

The V Digital Services proof page →

What is the complete methodology behind each KPI?

Published. Each metric carries its exact formula, measurement protocol, and the date it was frozen, on the methodology page. Methodology changes get a new dated page; the old one stays live for audit.

Methodology →Transparency →

Do clients get raw, exportable before/after prompt outputs — not just dashboards?

Yes. Engagements produce raw prompt transcripts and receipts, and they persist. The V Digital Services baseline alone wrote 30 receipt files — 10 screenshots, 10 raw API responses, 10 text transcripts, each raw response hashed. The receipts page publishes signed artifacts anyone can verify without our cooperation.

Receipts →

How is the prompt panel constructed?

Fixed wording, named engines, dated runs, repeated over time — pulled through the API from a data-center droplet, with no login, no chat history, and no personalization. When a prompt changes between runs, the published study says so rather than blending the results; the V Digital Services study carries exactly that disclosure.

What prevents the AI-facing content from diverging from the public site?

The machine layer is derived from the client’s approved public content — it restructures what is already true and published, it does not invent. Assertions are grounded before publication, and when origin content is retired, the corresponding machine-layer page is retired with it. Any URL’s AI-facing response is inspectable on request (see Question 01).

How does bot verification, routing, and rollback work?

AI crawlers are identified by user agent; human visitors always receive the origin site. Crawl telemetry is recorded and shown to the client on their dashboard. Rollback is a routing change: the origin site is never modified, so ending an engagement leaves the site exactly as it was.

Who owns the AI-facing content and the telemetry?

The machine layer is built from the client’s own facts, and the client keeps it. Telemetry gathered during an engagement is export-available to the client. Termination procedure is stated in the engagement agreement before signature.

Can this be structured as a pilot with a success gate?

Yes. Engagements can be structured as a defined pilot: a pre-agreed baseline, a fixed prompt panel, success thresholds set in writing before the start, and no-renewal as the stated consequence of a miss.

No vendor, including AnswerShare, can guarantee a specific AI answer on a specific prompt. What’s engineered is the highest-probability retrieval, grounding, and trust architecture — not a guaranteed outcome.

If a vendor will not answer these questions in writing, that is also an answer. See the receipts.