About The Companion Office

The Companion Office tests AI companion apps under one versioned method and publishes the ranked conclusion first.

Scope

The publication covers AI companion apps discovered through a recurring market census across app stores, web-native products, recurring search results, material launches and moderated suggestions. Every product that completes the minimum evidence bundle is ranked under the same contract: one consecutive position, the same logo treatment, score, verdict, drawback, receipts and direct destination. There is no featured section and no paid position.

How the testing works

Each app is used for a minimum test window on a declared tier, with first-session, 24-hour, 7-day and 14-day checkpoints across a fixed scenario set. Results are stored as dated observations; scores are calculated from those stored results under a published, versioned method rather than written by hand. Country editions add a local overlay covering native-language performance, availability, billing currency and local pricing.

Limitations of this preview

This is a private staging build. The records behind every number are structurally complete but illustrative: they demonstrate how evidence is reported and must not be read as findings about the named products. Real dated test bundles, verified destinations, prices and logo permissions are required before anything is published or indexed.

Corrections

Reported errors enter a private review queue. An accepted change is published in the corrections log with the affected record, the old value, the new value, the reason, the date and whether a score moved. Nothing is published automatically and there is no public comment surface.

Contact

Write to office@companionoffice.com, or use the structured templates on the feedback page.