Six evaluation dimensions
Proactive companionship, interaction quality, memory and continuity, mobility and environmental fit, privacy and account control, and total cost of ownership. We adjust the weighting for each use case.
Methodology
How we judge whether a robot is worth living with
Our evaluations are not a count of features. They focus on whether the experience can remain useful and enjoyable in a real home.
Three evidence levels
Level A is repeated hands-on testing with a recorded protocol. Level B is remote verification through manuals, support records, demonstrations, and multiple credible independent observations. Level C is a documentation-based assessment using official pages and policies. Every review states which level it currently reaches.
Claims are not proof
We prioritize product pages, manuals, help documents, privacy policies, warranty terms, and service notices. A manufacturer's description is attributed as a claim until repeatable testing or reliable independent evidence supports it. The absence of a public policy or manual is recorded as uncertainty.
Future hands-on test protocol
When a review unit is available, we will record setup time, interaction success and failure counts, touch-response latency, recognition consistency, obstacle encounters, battery and charging behavior, offline behavior, app permissions, account deletion, and support response. Test photos must be our own and dated; promotional images will not be presented as test evidence.
Limits and updates
Software features, subscriptions, and regional availability change quickly. Every article shows a review date, and material feature or policy changes trigger a fresh review. We do not fabricate first-hand experience, measurements, owner quotes, or independent validation.