About Vwrap
Vwrap is built by a small independent team. The research index, the generation pipeline, the editor and the words on this page are all our own work, and support@vwrap.ai reaches us directly.
What Vwrap is
A YouTube research tool with a thumbnail generator attached. Vwrap indexes channels and videos, scores each video against its own channel's baseline rather than against a platform average, and lets you send a video that beat that baseline straight into the generator as a reference image. The editor then works on the result a layer at a time: replace a person, replace an element, change a camera angle or an expression.
Those three things sit in one product because they are one job. Working out what is winning in a niche and making a thumbnail for it is the same afternoon's work, and moving between a research tab, an image tool and an editor loses the thing you just found. The target user is a creator between roughly 1K and 200K subscribers who ships weekly or every other week.
The background that matters here is the product rather than a résumé: the corpus, the outlier ranking, the generation pipeline and this site are maintained by the same small team, which is also why the limits below are on the page instead of buried.
What it does not do
Long-form and Shorts baselines are never mixed, so a Short is only ever compared with that channel's other Shorts. That is a deliberate constraint, and it means a channel with too few videos in one format gets a low-confidence label rather than a confident-looking number.
Multi-layer prompt edits are not supported. You can prompt one layer at a time, or prompt the whole image, but not two layers in a single instruction. Diffusion editors handled that badly enough in testing that shipping it would have been worse than leaving it out. The editor canvas is fixed at 16:9, so an input that is not 16:9 is stretched to fill it.
How the pages on this site are written
They are drafted with AI assistance and reviewed by a human before publishing. That is the disclosure in full: a model produces a first draft, then it is cut, corrected and read end to end by us, and nothing goes up that a person here has not read as a page.
The numbers get a second gate, and this is the part worth knowing. Product figures on this site (credit cost per model, plan credit grants, edit prices, trial length) are not typed into the copy from memory. Each one carries the path to the constant it comes from in the application's own source code. An automated check runs on every build, reads those constants, and fails the build when a figure on the page and the constant behind it disagree. A stale number stops the deploy instead of shipping.
It verifies figures, not prose. The check has no opinion on whether a sentence is true and it cannot see a claim that contains no number, so a wrong sentence still gets through if a human misses it. If you find one, email support@vwrap.ai. Corrections are the fastest thing here to ship.
Contact
support@vwrap.ai for support, billing, privacy requests and press. Email is the only channel: there is no phone line and no live chat. The Terms of Service, Privacy Policy and Refund Policy carry the registered seller details, the named subprocessors and the data-protection contact.