If you want the short version: Vidy's Google Play listing sells “Real-time AI video conversations,” and its own privacy policy names four AI vendors — Gemini Live, Gemini, Qwen Omni and Doubao — every one of which it describes as handling text, audio or speech-to-text. There is no video-generation vendor in that table. That does not prove anything about what you will see on screen, but it is the single most checkable fact either company has published, and it is the one worth starting from.
This piece compares Vidy and Prelulu on what each company states in public, checked on 2026-09-02. Where something could not be verified without an account, it says so instead of guessing — which is the same standard we would want applied to us.
What each app publishes about itself
| Vidy | Prelulu | |
|---|---|---|
| Store listing claim | “Real-time AI video conversations” | Live, lip-synced video call |
| Named AI vendors | Gemini Live, Gemini, Qwen Omni, Doubao — all text / audio / speech-to-text | Own avatar model, rendered by us |
| Video-generation vendor named | None in the privacy policy's service table | n/a — the model is ours |
| Real-time transport | aRTC (named in the privacy policy) | LiveKit |
| Paid tiers | VIP $19.99, SVIP $29.99 (Apple listing) | Free · Plus · Premium · Patron |
| Free entry | “First-time users get a 7-day free trial” | Text chat free forever; a free minute of voice daily |
| Age rating | 18+ on Apple, “Everyone” on Google Play | Tasteful by design; every character is fiction |
The age-rating row is not a nitpick. The same app is rated 18+ by Apple and “Everyone” by Google Play. Whichever rating is correct, one of the two stores is showing the wrong one to parents.
The privacy policy is the most informative page Vidy has
Vidy's privacy policy contains a service-provider table — the list of third parties that touch your data and what each one does. Four rows are AI services: Gemini Live for generating audio-based AI responses, Gemini for text-based responses, Qwen Omni for processing voice input into voice responses, and Doubao for speech-to-text. A fifth row names aRTC for real-time audio and video transmission, and AWS for hosting.
Read that table as an ingredients list. Audio in, audio out, text in, text out, and a transport layer that carries video. What is absent is a vendor whose job is to make the video frames. A company that licensed a video-generation model would have to name it there, because that is what the table is for.
The one screenshot detail worth noticing
Vidy's second Google Play screenshot is captioned “Video Call” and shows the AI character filling the screen with three controls — microphone, camera and end call — and no self-view thumbnail of the user. Its Play Data safety page lists “Voice or sound recordings” and “Photos” as optional collected data types, and does not list Videos.
So the camera button is there and the data-safety page does not claim to collect video. Those two facts sit oddly together, and the honest reading is that we cannot tell from outside which way it resolves.
The security disclosure you should read before either app
Vidy's Google Play Data safety page declares “Data isn't encrypted” and “Your data isn't transferred over a secure connection.” Those are Google's own standard phrases, shown because the developer answered the questionnaire that way.
We are a competitor, so treat this as the interested party pointing at a public page rather than as advice: go read it yourself, on the store listing, before you send anything to any AI companion you would not want read back to you. That applies to us too.
What Prelulu actually is, stated plainly
Prelulu is an AI character-companion app. You browse characters, text chat with them free, and have a live video or voice call on an avatar model we own and render ourselves. Characters remember your story across conversations. Text chat is free forever with full memory and no per-message meter; the first live call is on us, and there is a free minute of voice every day. Live video and voice beyond that are paid — generous, but finite.
The mechanism is the comparison that matters, and it is one you can check in ten seconds of using either app: a model that renders a face per word, against a video pipeline whose vendors are not named.
How to judge a video claim in any AI companion app
- Say something with hard consonants — “pumpkin,” “bottle.” Watch whether the mouth shape changes per word or loops.
- Interrupt mid-sentence. A rendered face stops; a clip finishes its loop.
- Ask for a specific expression. Generated video can attempt it; a looped asset cannot.
- Read the privacy policy's service-provider table. It names who processes what.
- Check the store Data safety page — both stores, because they can disagree.
Who each one is for
- Vidy might suit you ifYou want an 18+ app with unlimited-session framing and you are comfortable with the security disclosure on its Play listing. Its Apple tiers are VIP $19.99 and SVIP $29.99.
- Prelulu might suit you ifYou want text chat that stays free with full memory, a character who remembers your story between conversations, and a live face you can check for yourself.
- Neither, ifYou want a general assistant. Both of these are companions; a companion app is a worse ChatGPT and ChatGPT is a worse companion.
How we checked, and what we could not
Every Vidy specific above was read on 2026-09-02 from a public page: the Google Play listing and its Data safety page, the Apple App Store listing, vidy.chat, and vidy.chat's privacy policy and terms. We did not create a Vidy account, so we have not seen a call, and we make no claim about what a Vidy call looks like.
We also could not find a price on vidy.chat: the site's navigation has no pricing entry and /pricing, /plans, /en/pricing and /en/plans all return 404. The prices above come from the store listings.
See a face render, then decide
Text chat is free forever and the first live call is on us. Watch the mouth shapes and judge for yourself.