NoLang video AI adds one‑click horizontal and vertical conversion

NoLang’s new format‑conversion feature is a small but telling example of where applied generative AI is heading: into the plumbing of everyday content operations. By automatically re‑composing shots, subtitles and on‑screen elements for vertical and horizontal formats, Mavericks is trying to turn what used to be tedious editing work into a one‑click operation. That doesn’t push the frontier of reasoning or multimodal understanding, but it does show how video‑native AI tools are maturing from novelty generators into workflow engines.([prtimes.jp](https://prtimes.jp/main/html/rd/p/000000069.000129953.html))

In the race to AGI, this sort of productization matters because it funds the ecosystem and creates real‑world feedback loops. A service with 150,000 users and dozens of enterprise customers—numbers NoLang cites—generates diverse usage data on where current models fail to preserve framing, legibility or narrative when formats change. Those failures feed back into improved scene understanding, layout planning and adaptive rendering, all useful competencies for more general video and agentic systems. It also signals that Japan’s homegrown generative AI companies are carving out niches in video and creative tooling rather than trying to clone US‑style chatbots.

As more of the world’s marketing, recruiting and education content becomes AI‑assembled video, the line between "production" and "inference" will blur. Companies like Mavericks are betting that whoever controls that layer will have outsized influence over how and where future multimodal foundation models are deployed.