She had the AI build a pre-publication audit assistant; the audit failed three times, the alt-text tool worked
A Japanese note writer (TOIRO) who publishes US special-education explainers for Japanese-speaking parents used AI to build a pre-publication audit assistant, after months of copy-pasting drafts into chatbots for fact, source and accessibility checks (her September 2026 Back-to-School Support Navigator PDF for parents needed repeated legal-accuracy checks). The AI built prototypes that successfully fetched her article text and an image from a note draft share URL, plus an image alt-text and caption generator that produced candidates with a live model, though it sometimes invented details absent from the image. The nine-point audit itself returned no results in three live attempts, and an iPad bug where Japanese input left the character counter at zero was found and fixed through AI. Offline tests pass (43); the live audit is still unverified.
How it works
- Delegate the build brief to the AI (prompt drafted with AI)
- The AI prototypes an alt-text checker
- Adds draft-URL fetching, which succeeds on a real draft
- Builds the audit UI
- IPad Japanese-input bug found and fixed via AI
- Nine-point audit implemented
- Three live audit attempts all return no results
Why it’s worth knowing
The agent built the checker, fetched the draft by URL and generated alt text live, yet the headline audit silently returned nothing three times in a row.
Where it’s from
Summarized from a public post on note.com. See the original ↗