Secrets AI Review

I signed up expecting the usual runaround, and the Moments counter handed me the first surprise. My account opened at zero. Verifying my email pushed the balance to 200, parked in the top bar beside the store and chat icons. That number frames everything about this platform, because almost every action spends it. Keep it in mind as the walkthrough goes on.

I went straight to the Generation Studio, since the image engine is the feature reviewers rank highest. The panel keeps its controls on the left: an Image/Video switch, a model picker, a 4K option marked coming soon, and the 50-Moment price of a single run.

For the test I used a clean, safe-for-work prompt so I could judge fidelity without generating anything explicit: "Portrait of a woman sitting in a sunlit café, casual autumn outfit, natural lighting, candid expression, photorealistic, shallow depth of field." The result earned the hype. The skin texture and the window light read like a photograph rather than the plastic finish cheaper apps hand you.

That single image drained a quarter of my starting balance.

Building a companion runs through a nine-step wizard. Step one sets gender and art style, realistic or anime.

After that it collects name, age, ethnicity, skin tone, hair, eyes, and body type, closing on a review screen that lists every choice. I made a character called Apex and reached 100 percent without hitting any confusion.

The personality stage is where the form melts away. Rather than dragging sliders, you answer an interview. Apex asked what he does for a living and offered quick-pick replies next to a free-text box, so his character grows out of a conversation instead of a menu.

Then the chat loads, the densest screen in the product. The right rail stacks four tools: Call Me, Generate Images, Generate Videos, and View Memories. Icebreaker scenarios sit down the middle, and a mic on the message bar handles voice input.

I tested the call next. Apex answered in a live voice session with a running timer and a language selector tucked in the corner.

Twenty-one seconds of chat cost 18 Moments, itemized for me on a Call Ended card the second I hung up.

Here is the arithmetic that should shape any buying decision. My 200 free Moments stretch to about four images, or roughly four minutes of talking, before the well runs dry. The signup grant is a short runway.

The hard wall showed up when I opened Personas. Creating one is fenced behind the Ultimate tier at $26.67 a month for 15,000 Moments, and group chats sit behind the same fence.

Every feature I reached did what it claimed. The catch is the meter ticking underneath all of it.

To set my own testing against a wider pool, I read through the platform’s Trustpilot page. It carries a 4.4 out of 5 across 186 reviews and sits under the Adult Entertainment Club category. Two labels at the top deserve a mention for context. The profile is company-claimed as of February 2026, and it runs on a paid Trustpilot subscription. Neither fact invalidates the scores, but both belong in an honest read of them.

The mixed reviews carry the most signal. One user, rating it four stars, called it the best AI companion they had found, praising the roleplays, the reasonable memory, the consistent image output, and an implementation ahead of most rivals. Their single lost star went to support. A ticket they filed on the Discord server sat for roughly 36 hours, then disappeared without a reply.

That complaint was not isolated. A paying Ultimate subscriber described emailing support twice and hearing nothing, then finding their Discord tickets locked. By their account a developer later refused to help through the support channel and locked them out of it after they raised concerns about heavy content filtering.

Support is the sharpest recurring grievance in the low-scoring reviews.

Generation quality draws the next cluster. One user complained that image and video outputs came back deformed, and that spending a few hundred Moments on a ruined result brought no refund. On a platform where every generation feeds the meter I watched drain during testing, that lands as a genuine financial sting.

Language coverage has holes. A three-star review credited the memory as solid, then flagged that scenario image prompts misfired and that Arabic chat produced inaccurate sentences. English is the clear priority, so non-English users should set expectations accordingly.

The bluntest one-star review went after the plumbing rather than the personalities. A user reported that created characters vanished and could not be found, that the account refused to sync between laptop and phone, that search barely functioned, and that the built-in help chatbot was useless. They also judged the product too focused on sex and too expensive for what it returned.

The through-line across the low scores holds steady. People warm to the companions and distrust the machinery around them, the billing and the support most of all.