A few months ago I wanted a 12-page printable guide for a workshop. The choices were ugly: open InDesign and lose two days, drop content into Canva and fight templates, or paste everything into ChatGPT and copy the result back into Word.
That gap is what got me building Booklet AI. The idea is simple: describe what you want in chat, the AI asks clarifying questions about audience and tone, gathers material from the web and any files you upload, generates an outline you can tweak, then writes and lays out 8 to 15 A4 pages with images and text mixed in a magazine-style flow.
Three things I learned while building it:
- Layout is the hard part, not text. Anyone can produce 1500 words; making them fit a fixed A4 grid with figures, captions, and section breaks is where most AI tools quit. Generating each page as a self-contained HTML block instead of streaming Markdown made the layout problem tractable.
- Source quality beats prompt cleverness. Letting users upload their own PDF, Word, or image and feeding that into the model gave a much bigger lift than tweaking system prompts.
- Edit modes matter. Two flows on every page (natural-language AI Edit and visual drag Quick Edit) cover both kinds of users — those who can describe what they want and those who only know it when they see it.
Free to try at bookletai.org. Curious to hear what people use it for.
Top comments (0)