🟡 Weekend — A focused Saturday + Sundaynext.jsai-sdkvision
Images across a site with no alt text, and writing it by hand is a slog
“Accessibility audit flagged 200 images with no alt text. Writing all of them manually is soul-crushing, so it never gets done and screen-reader users suffer.”— Hacker News ↗
The gap
Vision models can describe images, but there's no simple batch workflow tuned for *alt text* specifically (concise, contextual, decorative-vs-meaningful). A drop-in-images → review → export tool fills a real compliance gap.
For
Web devs and content teams chasing accessibility compliance
Paste-ready build prompt
Build a bulk alt-text generator in Next.js. User drops in multiple images (or pastes URLs); a vision model drafts concise, contextual alt text for each, with a flag for 'likely decorative'. User edits inline and exports as a list/JSON. First 3 steps: (1) multi-image upload grid, (2) per-image vision call returning `{alt, decorative}`, (3) an editable review table with copy/export. Emphasize concise, non-redundant alt text in the prompt.