bestgirlfriend.ai
Best Match Quiz
About

About bestgirlfriend.ai: Methodology, Score-Lock, Errata

Meet Alexandra Joly, senior editor. We test AI companions, cam sites, real creators, adult games on transparent rubrics. Scores lock before commissions.

By Last revised May 2026

I started bestgirlfriend.ai because the AI girlfriend, cam, adult-games, and creator-subscription spaces are dominated by anonymous review sites with fabricated credentials, undated pages, and rankings indistinguishable from advertisements. Three consumer publications I read closely in adjacent categories (Wirecutter, Consumer Reports, and ProPublica) built reader trust on the same three choices: name the editor, publish the method, log every correction. So that is what this site does, adapted to a category where the incumbents have done the opposite.

This page introduces me, the scoring frameworks I run, the structural firewall between affiliate revenue and editorial scoring, and how to contact me when something on the site is wrong. The companion trust pages are our scoring frameworks, the editorial workflow that orders scoring before commission, the affiliate disclosure covering the commercial half of the contract, and the corrections log.

Who runs bestgirlfriend.ai?

Alexandra Joly runs bestgirlfriend.ai. She is a 32-year-old Paris-based editor who has tested AI girlfriend apps, cam sites, real-model creators, and adult games since January 2026. The site is independent, reader-supported through CrakRevenue affiliate links, and not owned by any platform she covers. She signs every review and locks every score before the payout is consulted.

Hi. I'm Alexandra. I'm the editor, the lead tester, and the person on the other end of [email protected]. There's no committee of anonymous staff writers behind me, no fictional "editorial board" carrying my byline as cover. The human team is small on purpose. I'd rather publish one review I've tested first-hand than ten I haven't.

The publisher entity holds the CrakRevenue affiliate account, the domain, the VPS, and the mail relay. I'm salaried; I'm not paid per click, per conversion, or per ranking position. The Schema.org Organization node on the homepage and the Person node on this page link back here on every byline. The structural separation between the publisher entity and what ends up in a review is documented in full on our editorial process page.

Is bestgirlfriend.ai's content written by AI?

The testing and the scores are mine. The drafting and the translation into the other languages we run in are done with AI tools, then edited and fact-checked by me before anything goes live. No score on this site was generated by a model. Every score came from a test I ran.

I'm not going to pretend otherwise. AI turns my test notes into readable English and into the languages we publish in, then I read every line back before it ships. What it never does is the part that matters: it hasn't used a companion for a week to watch the memory drift, and it hasn't seen a charge land on a card statement under a name it didn't recognize. I have. The tool assists. It doesn't test, and it doesn't decide.

Who is Alexandra Joly?

Alexandra Joly is the senior editor of bestgirlfriend.ai, in role since January 2026. She holds a Licence Information et Communication from Université Sorbonne Nouvelle (Paris, 2013-2016) and worked as a freelance journalist on tech and digital adult industries from 2022 to 2025. She is bisexual, which is the operational reason one editor can credibly cover both girlfriend and boyfriend modes, both straight and gay cam, and both women and men creators. Contact: [email protected].

I'm 32, born September 1994, French, based in Paris. Three years of public-university journalism school at Université Sorbonne Nouvelle (Licence Information et Communication, 2013-2016), then a slow build through SEO and copywriting work from 2015, then a freelance journalism slot covering tech and adult-industry stories from September 2022 through end-2025 (technology, consumer software, digital adult industries). Senior editor here since January 2026.

The dominant pattern in this space is fabricated bylines. One well-resourced competitor in our category claims a "Coursera Bachelor 2005-2009" credential on its About page, which is a credential that can't exist (Coursera was founded in 2012). Another lists "expert reviewers" with stock-photo headshots, nonexistent universities, and zero professional traceability. So I went the other way: real degree from a real public university, real LinkedIn, real Bluesky with DNS verification on this domain, real mailbox I read personally.

You can verify me on LinkedIn, on Bluesky (the handle binds via a TXT record on bestgirlfriend.ai DNS, which is cryptographic proof I control the domain), on X, and on my Substack where I publish monthly editorial notes. Same face on every profile. Same email. Same byline.

Why is bisexual coverage important here?

Because the site covers AI girlfriend AND AI boyfriend, straight cam AND gay cam, women creators AND men creators across 12 product categories. Alexandra tests both modes of every dual-orientation app and watches both sides of every dual-orientation cam platform. The bi part is the structural answer to "how does one editor credibly score products built for different audiences?", it is not a diversity statement, it is the testing protocol.

Most review sites in this space stay on one orientation, either by design or by hidden default. They write "best AI girlfriend" pages with male users in mind, ignore the AI boyfriend variant of the same app, and either skip the gay cam half of the cam category entirely or hand it off to an outsourced contributor who never tests it. The lopsided coverage shows up in the reviews, even when the site doesn't say so out loud.

I do not have that gap because I am bi, and because I find the products I am scoring interesting from both sides. When I run the Candy.ai test, I build a woman persona one session and a man persona the next, customized down to the kink I am in the mood for that week, and I score both runs. When I review a cam platform, I open the straight rooms and the gay rooms during the same peak-time observation window. When I cover real-model creators, I subscribe to women on OnlyFans and men on JustForFans in the same audit cycle. The pronoun alternation in my reviews (her (or him), a girl or a guy, the persona goes where you take her) is just the natural way I write, because that is the actual product I am using.

The structural payoff: one named, verifiable editor signs the AI girlfriend Pillar, the AI boyfriend Pillar, the gay AI chat coverage, the straight cam rankings, the gay cam rankings, the women-creators leaderboards, and the men-creators leaderboards. No invented second persona, no anonymous gay-specialist byline, no editorial trick to plug a coverage gap.

How does bestgirlfriend.ai score products?

Four parallel scoring frameworks, one per product category. AI companion apps are scored on 8 categories (conversation, image gen, customization, privacy, pricing, UX, voice, video). Cam sites on 6 (model variety, pricing & tipping, broadcast quality, payment & geo, privacy, UX), with a $0-spend testing protocol. Adult games on 7 (including a Billing Transparency axis nobody else publishes). Real-model creators on 6 (content volume, engagement, pricing, audience match, privacy, production quality). All four are published in full on our scoring page.

The four frameworks aren't interchangeable. An AI girlfriend app and a cam site are different products and the categories that matter differ; trying to grade them on one unified scale would either pretend they're the same product (they aren't) or hide whichever axes are inconvenient. So there are four. Each one is published with its dimension weights, its testing protocol, its re-test cadence, and its version history.

The AI companion framework weighs Pricing & Value at 18%, Conversation at 16%, Privacy & Compliance at 14%, Customization and Image Gen at 12% each, UX and Voice at 10% each, and Video Gen at 8%. I run a 10-prompt conversation, a 5-image generation set, and a standardized voice phrase on every platform I test, then peer-review before publish.

The cam-site framework weighs Model Variety & Volume and Pricing & Tipping at 18% each, and Broadcast Quality, Payment & Geo, Privacy & Compliance, and UX & Mobile at 16% each. The $0-spend protocol means I observe peak-time inventory across three timezones, run a 10-room broadcast quality test, and walk the pricing page and checkout flow up to (but never past) payment submission. I do not charge a card to score a cam site. Post-purchase reality (billing descriptors, refund friction, auto-renewal triggers) is sourced from aggregated user reports and flagged as not independently verified by me when I cannot test it directly.

The adult-games framework has a unique 7th axis nobody else publishes: Billing Transparency. Three years of Trustpilot complaints, scam-detector flags, and Reddit threads on the Kinkoid-portfolio brands (Hentai Heroes, Harem Villa, Pornstar Harem, and siblings) document a pattern of opaque auto-renewal, browser-redirect malware history, and refund friction that the headline game-quality score does not capture. So we score it explicitly.

The real-models framework weighs Content Volume & Cadence, Engagement & Interaction, and Pricing & Value at 18% each, then Audience Match at 16%, Privacy & Compliance at 14%, and Production Quality at 16%. Per-creator scoring is harder than per-platform scoring because the data window is shorter (one creator's cadence changes faster than one platform's pricing page), so I re-audit creator pages more often than I re-audit platforms.

All four frameworks are versioned. When a major change ships, every affected page is re-scored within 90 days, with the reason documented and the page's revision date updated to match.

How does the score-lock work?

Scores are computed and locked before the CrakRevenue payout for the platform is consulted. The scoring and the score lock both come before the commission cross-check in our editorial workflow, so a score is final before the commission is even known. After publish, the score does not move without a documented re-test, and the page's revision date updates whenever it does.

The point of locking scores up front is so the affiliate commission can't talk us into a 7 when the reality is a 5. Most reviewers in this space won't say that out loud. They polish the rough bits on the platforms that pay well, skip the inconvenient ones, and bury the dimensions where the high-commission product underperforms. We don't.

Concrete example: the Joi.ai review pays one of the highest commission rates of any AI-companion app we review. It ranks fourth on our list, not first, because its image generation is weaker than Candy.ai's and its UX has friction Candy doesn't. The commission didn't move the score. Candy.ai's review (8.4) pays a mid-tier commission and ranks first on image polish, fourth on memory, with Memory scored 5/10 not 7/10 because conversations go fuzzy around day five and I tested that twice. Reddit threads converge on the same complaint. The 5/10 stays.

The three structural defenses against editorial capture (process firewall, documentation firewall, audit firewall) are detailed on the editorial process page. Each one alone is bypassable. Stacked together, they make rigging a score more work than just reporting the real one.

How does bestgirlfriend.ai handle affiliate commissions?

When a reader clicks an outbound link tagged rel='sponsored' and signs up or pays on the destination platform, CrakRevenue pays a commission to the publisher entity. Editors receive a flat salary, never per-click or per-conversion bonuses. Scores lock before the offer list is cross-checked, so commission rates cannot reach the scoring. Our affiliate disclosure page lays out the click-attribution format, the per-page disclosure block, and the FTC 16 CFR Part 255 compliance framework in full.

I won't pretend the site is free of commercial interest. It's an affiliate site. The commercial relationship is the standard CrakRevenue affiliate program, the same network that pays most other publishers in this space. The difference here isn't the absence of commission. It's that the commission can't reach the scoring step. The full architecture (the rel="sponsored" qualifier on every commercial link per Google's outbound-link guidance, the click-attribution format we use, the score-lock framework, and the per-page disclosure block required by 16 C.F.R. Part 255) lives on our affiliate disclosure page.

The methodology pages are the editorial half of the contract. The disclosure page is the commercial half. Neither half is hidden, and neither half pretends the other doesn't exist.

What does bestgirlfriend.ai cover?

Four product categories across 12 content areas. AI companion apps (girlfriend, boyfriend, anime, image generation, video generation, voice, uncensored chat), live cam platforms (straight default plus gay cam coverage), real-model creators (per-creator OnlyFans, Fanvue, SextPanther pages), and adult games (porn games, hentai games, harem games). Each area has its own scoring framework, its own testing protocol, and its own re-test cadence.

The site is organized into twelve content areas. Seven cover AI companions: AI girlfriend, AI boyfriend, AI anime and waifu, AI image generation, AI video generation, uncensored AI chat, AI companion safety, plus a cross-cutting alternatives and comparisons area. Two cover the cam world: straight cam (default) and a gay cam modifier landing. Two cross-cutting areas handle the AI-versus-cam bridge content and per-creator real-model coverage. The twelfth area covers adult games.

What bestgirlfriend.ai covers, by category and current scope
Last reviewed: 2026. Frameworks: /methodology.

How is bestgirlfriend.ai different from competitors?

Three structural differences. A named editor with a verifiable journalist profile (most competitors use anonymous bylines or fabricated credentials, including one well-known site claiming a Coursera credential from 2005, four years before Coursera existed). Four published scoring frameworks with version history (most competitors show a single hidden review process). And a public errata board logging every correction with its date and reason, where most sites silently edit pages or never correct at all.

The competitor landscape in this space includes one well-resourced affiliate site claiming a "Coursera Bachelor 2005-2009" credential that can't exist (the platform launched 2012). Several anonymous-byline directories publish "expert reviews" with no editor named, no testing protocol disclosed, and rankings that change weekly in lockstep with payout changes. The most common pattern: undated review pages with broken outbound links pointing at expired offers, and a single "How we review" page that says "we test every product" without saying who tested it, when, or how.

None of them publish a method, a re-test cadence, or an errata log. My position is that the absence of those three artifacts is the diagnostic, not the methodology page itself. If a publication won't say who tested the product and when, the score is unfalsifiable. If the page carries no last-updated date, the score is undated. And if there's no public errata log, there's no way to tell whether the publication corrects its mistakes or pretends they never happened.

I publish all three because that's what makes a review checkable. Whether you trust the score is up to you. Whether the score is checkable is up to me.

Editorial independence

bestgirlfriend.ai operates under three structural defenses against editorial capture. They aren't aspirational principles; they're testable rules embedded in the workflow.

Process firewall. Scoring and score lock both come before the CrakRevenue commission cross-check in our editorial workflow. A score is final before commission is consulted. The ordering is enforced by tooling, not by honor system.

Documentation firewall. A published score only moves on a documented re-test, and the page's revision date updates when it does. Where a score moves, the trigger event (regulatory action, terms-of-service overhaul, model swap, price change, breach disclosure) is named in the review itself.

Audit firewall. Weekly automated checks compare the freshness of every outbound offer link against the live offer feed and flag any commission-rate change that isn't paired with a documented re-test. A human reviews each flag. If a re-test is warranted, the affected dimension gets re-tested under the published cadence and the new score is logged whether it moves up, down, or stays flat.

Each defense is bypassable alone. Stacked together, they make gaming a score more work than it's worth.

How can I contact bestgirlfriend.ai?

Email [email protected] for corrections, errata, journalist inquiries, or platform contestation. Acknowledgment within two business days. The full mailbox routing (five purpose-built addresses with published service-level commitments) lives on our contact page: privacy@ for personal-data requests, dmca@ for copyright notices, accessibility@ for accessibility barriers, contact@ for everything else. Cold sales pitches are not responded to.

I read [email protected] personally. If a fact is wrong in a review, if a price has changed, if a platform you operate believes the score is unjust, that is the address. Acknowledgment within two business days, resolution per the per-mailbox commitments published at our contact page. Verified journalists, attorneys, and platform representatives can reach me on LinkedIn for matters requiring direct response within five business days.

How can I pitch a story?

Email [email protected] with subject line 'Pitch:' followed by a one-line headline. Include angle, named sources, and proposed length. Alexandra Joly reviews pitches weekly and accepts under three percent of inbound. Acknowledgment confirms receipt only, not editorial interest. Pitches offering payment for placement are discarded under the paid-placement policy.

The pitches I want are the ones I cannot write myself: primary-source access I do not have, regulatory filings, internal documents from platforms covered on the site, model performer testimony, payment-processor source documents on chargeback patterns, academic research on harm and consent dynamics in AI companion products. The pitches I do not want are re-aggregated trend pieces, opinion essays without sourcing, and AI-generated copy. Accepted pitches include a kill fee and a named byline; ghost-written content is not bought here.

Where is bestgirlfriend.ai based?

The publication operates as a small remote team. Alexandra Joly is based in Paris, France. A physical office address is not published publicly for staff-safety reasons in an adult category that attracts inbound harassment. Service of process and statutory correspondence (DMCA, GDPR Article 27 representative inquiries) is initiated by email through our contact page, with a confidential mailing address provided to verified counsel.

The remote structure is a function of the category, not a cost-cutting choice. AI companion, cam, and adult-games coverage attracts inbound harassment, and concentrating editors in a single physical office creates a single point of operational risk. Service of process and statutory correspondence are handled through the published email addresses, with a confidential postal channel for verified counsel. ProPublica and other publications operating under similar threat profiles use the same mechanism.

Cite this page

Joly, A. (2026). About bestgirlfriend.ai: Alexandra Joly, Mission & Editorial Team. bestgirlfriend.ai. https://bestgirlfriend.ai/about

Sources

Frequently asked questions

Who runs bestgirlfriend.ai?

Alexandra Joly runs bestgirlfriend.ai. She is a 32-year-old Paris-based editor who has tested AI girlfriend apps, cam sites, real-model creators, and adult games since January 2026. The site is independent, reader-supported through CrakRevenue affiliate links, and not owned by any platform she covers. She signs every review, locks every score before the affiliate payout is consulted, and posts every correction to a public errata board.

Who is Alexandra Joly?

Alexandra Joly is the senior editor of bestgirlfriend.ai, in role since January 2026. She holds a Licence Information et Communication from Université Sorbonne Nouvelle (Paris, 2013-2016) and worked as a freelance journalist on tech and digital adult industries from September 2022 to December 2025. She is bisexual, which is the operational reason one editor can credibly cover both girlfriend and boyfriend modes, both straight and gay cam, and both women and men creators. Contact: [email protected].

Why trust bestgirlfriend.ai scores?

Three reasons. Alexandra tests every platform before scoring it (10-prompt conversation for AI, peak-time observation for cam, free-tier playability for games, content-cadence audit for creators). Scores are locked before the affiliate payout is checked, so commission cannot lift a 5 to a 7. Every correction is logged at /errata with date, page, and reason. No paid placements, no sponsored reviews, no ranking-lift offers.

Is bestgirlfriend.ai's content written by AI?

The testing and the scores are mine. The drafting and the translation into other languages are done with AI tools, then edited and fact-checked by me before publish. No score on this site was generated by a model. Every score came from a test I ran. I disclose this because a solo-led publication producing in volume without help would not be credible, and concealing it would contradict the standard this site holds everyone else to.

Why does bestgirlfriend.ai cover both straight and gay content?

Because Alexandra is bisexual and tests both. The site covers AI girlfriend and AI boyfriend modes, gay AI chat, straight cam and gay cam, women creators and men creators across 12 product categories. Coverage parity is structural, not performative. The audience-isolation design keeps straight pages straight and gay pages gay so the experience matches the search intent, while the footer and sitemap expose every category to search engines.

What does bestgirlfriend.ai cover?

Four product categories: AI companion apps (girlfriend, boyfriend, anime, image generation, video generation, voice, uncensored chat), live cam platforms (straight and gay), real-model creators (per-creator OnlyFans, Fanvue, SextPanther pages), and adult games (porn games, hentai games, harem games). Each category has its own scoring framework and re-test cadence published at /methodology.

How does bestgirlfriend.ai make money?

Affiliate commissions from CrakRevenue when a reader clicks an outbound link tagged rel='sponsored' and signs up or pays on the destination platform. Alexandra receives a flat salary, never per-click or per-conversion bonuses. Scores are computed and locked before the affiliate offer list is cross-checked, so commission rates cannot reach the scoring. Full disclosure at /affiliate-disclosure.

Does bestgirlfriend.ai accept paid placements?

No. The site refuses paid placements, sponsored reviews, ranking-lift offers, link insertions, and guest posts. Brand emails proposing these arrangements are logged and ignored; persistent senders are blocked at the mail-server level. The only commercial relationship is the standard CrakRevenue affiliate commission, disclosed at /affiliate-disclosure and structurally separate from editorial scoring.

What is the score-floor rule?

No platform scoring below 5.0 composite is recommended on bestgirlfriend.ai, regardless of payout. A separate privacy-floor disqualifies any platform with a Privacy and Compliance sub-score below 5 from the recommended list entirely. Score-floor and privacy-floor are non-negotiable and audited weekly. Full mechanics at /editorial-process.

How can I contact bestgirlfriend.ai?

Email [email protected] for corrections, errata, journalist inquiries, or platform contestation. Acknowledgment within two business days. The full mailbox routing (privacy, DMCA, accessibility, general contact) is published at /contact with service-level commitments per address.

Where is bestgirlfriend.ai based?

Alexandra works from Paris, France. The publication operates as a small remote team and does not list a physical office address publicly, for staff-safety reasons in an adult category that attracts inbound harassment. Service of process and statutory correspondence (DMCA, GDPR Article 27 representative inquiries) is initiated by email, with a confidential mailing address provided to verified counsel.

This is the publication's About page. The companion trust pages (/methodology, /editorial-process, /affiliate-disclosure, /errata, /contact, and /privacy) together form the trust contract under which every review on bestgirlfriend.ai is published. Last reviewed: 2026.

About bestgirlfriend.ai: Methodology, Score-Lock, Errata