Who's actually running this

A site whose entire pitch is "evidence you can check" owes you a straight answer about who's doing the checking.

Independent, unfunded

CompanionsArena is not backed by any AI companion company, and no platform pays for placement or a favorable score. There is no revenue at all right now — no ads, no sponsorships and no affiliate commissions. If affiliate links are ever added, they will be labelled where they appear; see the affiliate disclosure for how they would be kept separate from scoring, in code, not just in policy.

The Admin — founder and lead tester

CompanionsArena is run by a single operator, referred to on this site as The Admin. Every platform review published under CompanionsArena's name has been personally tested and verified before publication — access type (free, paid, or provider-supplied) is always disclosed, and nothing goes live from a marketing claim alone.

How testing actually works right now

This is early. There is no testing team, no lab, no schedule beyond what one person and a small volunteer community can cover. Being upfront about that is more useful than pretending otherwise:

• Every platform runs the same public scenario pack — the exact prompts are published so you can run them yourself and compare.
• A platform is only scored on categories that have actually been tested. Where testing hasn't happened yet, the site says so rather than guessing — see the coverage note on any review page.
• Pricing, features and compliance facts that can be verified from a platform's own published pages (pricing page, terms, privacy policy) are researched and sourced even before hands-on conversation testing is complete — always cited, always dated, and clearly separate from tested conversation/memory scores.

The Discord community

A community of volunteer testers is forming on Discord. If you use one of these platforms and want to contribute a test run, transcript, or spot-check, you're welcome to join and send it in.

How volunteer submissions are handled

Anyone can run the public scenario pack and submit what they found. Nothing a volunteer submits is published automatically:

1. A submission includes the transcript, the plan tested, and the date.
2. The Admin reviews it — checking it against the scenario pack procedure and cross-referencing anything that seems off.
3. Only once verified does it become a published score with a rationale, same as a run done directly. Volunteer-sourced evidence is disclosed as such, not presented as identical to operator-run testing.

This mirrors the same principle that governs everything else on this site: a claim without a checkable source doesn't get published, no matter who it comes from.

Why this exists

Every AI companion review site claims rigorous, independent testing. None publish the evidence behind that claim — the test prompts, the raw transcripts, the arithmetic behind a score. CompanionsArena is built so publishing that evidence is the default, not an afterthought. See methodology for the whole architecture.

Get in touch

Corrections, tips, or volunteer testing: contact The Admin. See also the corrections policy for how disputed facts get resolved.