We selected four distinct brands from saved Muse and Instinct recommendations for first-time cosmetic Botox consultations in Dallas, Plano and Frisco. Fo and Kat each received one named-business inspection per brand. Eight assignments, all submitted on September 30, 2026.
The inspection asked for physical location, the named provider’s advertised qualification, treatment-specific prices and the public booking path. Agents could navigate public pages but could not enter personal information, submit forms, contact a clinic, reserve a time or pay.
Separate operator captures of four first-party pages were saved at approximately 16:48 UTC on September 30. We subsequently checked FRZN’s official FAQ. Captures are page text, not recordings of the assistants’ original tool activity.
These are selected cases, not a random sample. Agreement between assistants is not independent factual verification. Prior context and different tool capabilities can affect the observations.
Live-slot observations remain agent-reported and may change. A blocked fetch or a calendar an agent did not open is not evidence of a broken merchant system. Later message-based clarifications are kept separate from the original submissions.
We have not measured clinical quality, professional license accuracy, human-versus-agent differences, stable ranking factors, completed bookings or revenue lift. The supported conclusion is narrower. Service matching and appointment-path evidence deserve checks beyond the recommendation itself.
