Picture books tell stories through images, but writing about their illustrations was scattered and scarce in Chinese. Working solo as designer, developer and editor, I built a Traditional Chinese picture-book museum that connects historical context, visual analysis and sources.
To sustain that experience, I also built an AI research workspace inside the admin, with source-linked drafts and human review before publication.
AI Studio gathers evidence and drafts inside the admin, so one curator can keep publishing.
33 of 71 books are published, each with sourced visual analysis.
A review queue keeps a human approval step on each paragraph.
I wanted to understand how the pictures tell the story. Interviews, exhibition pages and articles were scattered across platforms, Chinese-language writing on illustration was limited, and references I saved on Facebook and Instagram were hard to find again.
I built the museum to bring historical context, visual analysis and sources into one reading experience.
Each entry focuses on how the illustrations tell the story. For Alice in Wonderland, I explored how wood engraving creates texture and depth through line. Cover-based palettes carry each book’s visual character into the page.
Testing with five readers highlighted a need for more visual analysis and help discussing books with children. I added close-read mode, focusing each illustration on three or four selected details. Readers can jump to an annotation, zoom freely and move between observations without finding each marker again.
Producing each sourced close reading required research, visual study, writing and fact-checking—about 7–11 hours per book. With 38 books waiting and a team of one, I needed a way to keep publishing without giving up that level of care.
Each field was written in isolation, so AI produced generic commentary about friendship and childhood instead of analyzing the illustrations. I still had to find sources and refocus the draft on visual storytelling.
Analysis of illustration and page layout
General commentary, not visual analysis
I split the work between a curation advisor in the admin and a writing skill in Cursor. The research improved, but drafts still lived outside the admin, and I copied them back by hand.
I merged both tools into AI Studio inside the admin. AI gathers evidence and drafts with the book’s context and sources available in one workspace, reducing manual transfers between tools.
AI-generated paragraphs link to numbered evidence, including sources and visual observations. I can open the evidence beside a paragraph, check the claim and review the text before approving it for publication.
The museum has 33 published books, each with a sourced visual close reading. AI Studio brings research and drafting into the admin, while paragraph-level review keeps publication decisions with me.
Producing an entry now takes about 2–3 hours, compared with 7–11 hours before.
Visit the museumThe useful shift was from generating isolated fields to supporting the whole editorial workflow. Shared context and fewer manual transfers made the process more workable; visible evidence and human review kept the final judgment with me.