Safe preflight
We normalize URLs, reject private network targets, cap redirects and response size, check product and required-media reachability, collect metadata, inspect public GitHub repository state, and flag duplicates.
Correction desk
Keep the directory
accurate.
Send the listing URL, the fact that needs attention, and a source. We review corrections under the same evidence standard as new products.
Review policy
Automation gathers.
Editors decide.
RetivAI separates a product’s availability, claimed technology, realtime mechanism, and evidence strength. A check is a lead—not a verdict.
Editorial review
A reviewer confirms that the product exists, qualifies as real-time AI video, has adequate evidence, uses a precise mechanism label, and can be described with original factual copy.
Weekly signal
Availability checks log failures. Two consecutive failures create a review flag, never an automatic deletion. A 200 response never proves the experience still works.
We can include
- Native continuous generation sessions
- Buffered or just-in-time interactive video
- Synthetic livestreams with a credible generation loop
- Developer infrastructure specifically enabling these products
We normally decline
- Ordinary offline text-to-video generators
- Marketing pages with no public evidence
- Duplicate or unsafe destinations
- Misleading, rights-infringing, or primarily harmful products
Evidence labels
- Official: first-party product or API source
- Source-code confirmed: public implementation
- Creator-confirmed: attributable creator statement
- Third-party reported / unverified: clearly disclosed
Submission is free during launch. Inclusion, wording, classification, timing, correction, and removal remain at RetivAI’s editorial discretion. See the full terms and privacy notice.