HoneyPot FAQ

Two criteria pulled the average down. Media features score 7, because voice and image generation are paid-tier only and neither is a standout when unlocked. Privacy posture scores 8 rather than 9: the absence of a social layer is a real structural advantage, but the browser-only model shifts part of the exposure onto the device you use. Conversation quality and persona control both score 9, and those are what the product is actually built around.

It could change the memory number. Fourteen days plus a seven-day gap is enough to detect drift and to confirm detail-level recall, but not enough to judge whether continuity holds over months. Nothing in the sessions suggested a cliff. If we extend the window and the finding moves, the score and the date stamp on this page move with it.

No, and the two should not be averaged. The member rating aggregates ratings from users of the service and reflects satisfaction. The editorial score is one desk applying one rubric across a comparison set. They are close here, which is mildly reassuring, but they measure different things and the rubric is the one this page defends.

Three things, in order of how much they matter. Deeper long-term recall is the one that changes daily use, because free-tier memory thins after several sessions. Voice comes second. Image generation is third and is not the reason to upgrade here. Everything needed to create, edit and converse with a persona is available before paying.

Yes, on the free tier as well. Tone, pacing and interests are all editable, and edits apply from the next reply rather than requiring a new character. In our sessions this was the practical way to work: set an approximate persona, run a few exchanges, then adjust pacing once you can see how it answers.

Anything involving real people or their likenesses, anything involving minors, and anything framed as non-consensual. These limits are enforced by the service rather than left to the user, and they are documented rather than discovered. This review treats that as correct design. If a companion app is vague about where its limits sit, that vagueness is the finding.

They stay attached to the account. There is no feed, no public profile and no shared library, so a conversation has one reader by default. Deletion is handled on request through the operator's contact route, and the operator's own privacy policy is the authoritative statement on retention — not this page.

Age confirmation is required before access, and the service is adult-only. Every companion is a fictional AI character. That framing is stated by the operator and applies regardless of how a persona is configured, which is the answer to the question people usually ask next.