Emma Clarke

Emma Clarke is the pen name of AI Companion Atlas’s editorial team. We publish under a pen name for privacy reasons — this category attracts a level of personal attention we’d rather avoid — but everything published here is written by a real person, and every app we review is one we’ve opened an account on and used ourselves.

We think you deserve to know that up front. Plenty of sites in this space invent a reviewer, staple a stock photo to a fabricated résumé, and hope nobody checks. We’d rather tell you what Emma is and let the testing speak for itself.

What we actually do

We open a real account on every app and use it before we write a word. We build a companion, hold the conversations, push the safety limits, read the pricing screen, the billing descriptor and the cancellation flow, and go through the terms instead of the marketing page.

And we tell you where our test stops. Right now that’s the paywall: most of these reviews are written from the free tier, which is where the majority of readers start too. Any criterion we could only assess from a platform’s own documentation is capped at 8.0 and the review is marked provisional until we’ve been through the tier above.

That cap is the point. Anyone can claim to have paid for everything; nobody can check. A score with a stated ceiling tells you exactly how much of the product we saw. Those scores are revised in both directions once we go back — several of them will go down.

To date we’ve reviewed 19 apps. No press accounts, no comped access, no early looks.

Every one of them is scored on the same nine criteria, with the same weights. Here is what counts, and by how much:

The full method, criterion by criterion — including how a capped criterion is scored — is in our testing methodology.

What we don’t do

We don’t review an app we haven’t used. We don’t score from a features page or a press kit, and we don’t report a price we haven’t read on the platform’s own screen — where one is hidden until sign-up, we say so and leave the cell empty rather than quote someone else’s blog post.

We don’t rank by commission. Some of the apps here pay us if you sign up through our links — that’s how the site stays free, and it’s disclosed on every page where it applies. Some of the most important apps in this category run no affiliate program at all and will never send us a cent. They belong in the rankings anyway, because a guide that only covers what it earns from isn’t a guide.

And when a paying partner disappoints, we say so. If our scores never moved, they wouldn’t be worth reading.

What we can’t tell you

We test as adults using these apps the way most people do — for conversation, companionship, roleplay. We’re not clinicians. When we write about loneliness or attachment, we report what the research says and link to it; we don’t diagnose, and we don’t tell you what any of this means for you specifically.

We also can’t test everything. Long-term effects, what happens to your data after a company is acquired, whether a small studio will still exist in two years — we flag the risk when we see it, but we don’t pretend to know.

Corrections

These apps change constantly — pricing, models, content policies, sometimes overnight. Every review carries a last-verified date. If something we’ve published is out of date or wrong, tell us and we’ll fix it and note the change.