Editorial Policy

How we test AI companion platforms, who writes our reviews, and how we disclose affiliate relationships.

Testing Methodology

Every review is based on hands-on testing for at least 14 days. For each platform, the reviewer assigned to that language: (1) creates an account from a clean browser profile in their country; (2) sends a minimum of 200 messages across three character archetypes — typically one anime, one realistic, one fantasy or genre-specific; (3) tests memory by mentioning a specific detail on day 1, returning on day 7 and day 14 to check whether the AI recalls it; (4) tests voice generation (where available) by saving raw audio samples; (5) tests photo generation against a fixed prompt list to compare consistency; (6) documents response times, character drift, content filter behavior, and any errors encountered. Screenshots are saved for every claim. No claim makes it into a review without a screenshot or saved log to back it up.

Who Writes Our Reviews

Each language has a designated lead reviewer who writes for their market. Reviewers are independent contributors — they own their bylines, their methodology, and their conclusions. They are not required to recommend HoneyChat, and several reviews on our own blog rate competitors above HoneyChat for specific use cases. Reviewers disclose any prior relationship with platforms they test. See the full author list with credentials, social profiles, and language coverage at the bottom of this page.

Affiliate Disclosure

HoneyChat is the platform that pays our reviewers and hosts this blog. Every link to honeychat.bot or @HoneyChatAIBot in our articles is a first-party recommendation — we have a direct interest in your trial. Links to competitor platforms (Character.AI, Replika, SpicyChat, Janitor AI, Crushon AI, Polybuzz, etc.) are non-affiliate — we earn nothing from clicks to them, but we link to them so you can verify our comparisons yourself. We never accept paid placements, sponsored reviews, or "review my product for $X" arrangements. If a competitor offers us money to write favorably, we decline and disclose the attempt publicly.

Update & Correction Policy

Reviews are re-tested every 60-90 days. When a platform pushes a major change (pricing update, content policy change, new feature, outage pattern), we update the relevant review within 14 days and bump the page's updated date with a real content change — not a date-only refresh. When we get something wrong, we correct it inline, mark the correction visibly (strikethrough + edit note), and date the change. We never silently rewrite history. Feedback to corrections@honeychat.bot — we read everything and reply within 5 business days.