The method · the sourcing · the money

About this site

Last updated 27 July 2026

I started this because I couldn't find it. Every list of AI companion apps I opened was written for someone else, tested from the other side, and mostly assembled by people who had clearly never opened the apps at all.

Why it exists

Search for the best AI girlfriend app and you get hundreds of reviews. Search for the best AI boyfriend and you get those same articles with the nouns swapped, plus a two-year-old Reddit thread that's still more useful than any of them.

These apps aren't symmetrical. A male persona is a different product, written differently, and it fails differently. Some apps that produce a convincing AI girlfriend produce an AI boyfriend who sounds like a walking cologne ad, and you'd never learn that from a review that only tested the other side. So I tested this side.

Who writes it

Erin Vaughn. I'm not a technology journalist and I didn't come to this as a hobbyist. I downloaded an AI companion app at 1 a.m. after a breakup, then couldn't stop wondering why every review of them was written for somebody who wasn't me.

I run this site alone. I write everything on it, I test everything on it, and I pay for everything on it.

How the apps are tested

Every app gets the same setup: the same backstory for me, the same rough character brief for him, daily conversations, and a minimum of five days. If an app earns a good week I keep going, and the ones at the top of the list ran a month or more.

Four questions decide where an app lands:

Every entry says how many days I used it, because a two-day opinion and a month-long one are not the same thing and most lists hide the difference.

How the research pages are sourced

The statistics pages follow a stricter rule than the reviews. Figures come from the organisation that collected them, which means peer-reviewed research, government agencies, established polling firms, companies' own results filings, or news reporting on those. Never from another blog's roundup, because that's where bad numbers come from and where they get laundered into fact.

Every statistic carries the date it was collected rather than the date it was published, and where two credible sources disagree I show both instead of picking whichever one sounds better.

How the site makes money

Affiliate links. If you sign up for something through one, I may earn a commission at no extra cost to you. That's the whole model, and the full version is on the disclosure page.

What that doesn't buy: placement, position, or a softer write-up. Two companies have offered money for a spot on the list and neither is on it.

Corrections

Apps change constantly, pricing changes faster, and research gets superseded. If something here is wrong or out of date, tell me and I'll check it and fix it. Corrections to research pages get noted with the date.

One honest note

These apps were good for me during a stretch when I didn't have much left over for people. They weren't a substitute for the friends I eventually called, and on the genuinely bad nights, no app was the right thing to be talking to. If you're using one to get through something heavy, keep a real person in the loop.