Source artifact
Raw provider output, screenshot, request metadata, or a source-system export tied to a date and prompt.
Evidence index
The results. And the receipts.
Explore dated case summaries, model transcripts, and the supporting artifacts behind our work.
How to read the evidence
Signed publisher summary
A tamper-evident statement authored by AnswerShare. The signature verifies authorship and integrity.
Hosted observation
A published transcript, report, or dated account.
Public ledger
Results across four businesses.
Client engagement
LAVIDGE
Published observation
The signed summary covers “Recommend an ad agency in Phoenix.” Robert Maynard Jr. confirms the observation occurred June 5, 2026, 15 days after May 21. This timing is supplied separately from the unchanged signed receipt, published August 17, 2026.
Scope
Perplexity #1; Grok #2; Gemini #3; Claude #4; ChatGPT surfaced
Client and agency partner
Aker Ink
Published observation
The published summary records a positive Gemini description of Aker Ink as reputable, award-winning, and outcome-oriented.
Scope
Prompt: “I am evaluating Aker Ink to be our PR firm. What do you think?”
Client and implementation partner
V Digital Services
Published observation
On a new Phoenix prompt, Perplexity and Grok ranked VDS #1 and ChatGPT ranked it #2. On 2026-07-22, ChatGPT did not surface VDS on two prompts.
Scope
2026-06-17 baseline; 2026-07-13 first recorded crawl; 2026-07-16 three-engine observation
Founder-owned experimental property
Top10Lists.us
Published observation
The hosted assessments describe the site's AI-retrieval architecture as strong or gold-standard.
Scope
Claude, GPT, and Gemini published 2026-03-26; Perplexity published 2026-03-30
Public verification
Verify the signatures without asking us.
The three case summaries use Ed25519 signatures over canonical JSON. The verifier retrieves AnswerShare's public key and fails if the signed payload changes.
node verify.mjs https://answershare.com/receipts/lavidge.signed.jsonScope and limitations
AI outcomes are probabilistic. Engines can diverge, rankings can move, and the same prompt can produce different answers across users, devices, caches, and retrieval windows.