Findings
What we found, including the clean results.
Observed behavior from dated, bounded tests. Failures, clean results, withheld cases, and limits remain in the same record.
A finding begins with a reproducible observation.
Entries come from our own verification sweeps and are anonymized where identification would add no public value. We publish a result because the capture was clear and the behavior reproduced—not because a business engaged us.
- Subject and surface tested
- What the system said
- What the available source supported
- Capture date, method, and limits
The clean results matter too.
A log showing only failures would be an advertisement. Clean, unresolved, and withheld results provide the denominator needed to understand what a sweep actually found.
July 13 DFW sweep
Thirteen businesses were scanned against one public AI-answer surface on one day. Four material findings reproduced, six results were clean, one returned no answer, and two were withheld because the available capture was ambiguous. This is a dated sweep—not a prevalence estimate.
Daily AI Agents