Moving along
I've developed a process for moving through the book, and made it through 9 chapters.
The map of Morris' travels, draft as of today. Map drawn by Claude.
I'm using Claude Opus 5.5 on the "Extra" effort setting. I paste in the OCR text, and have the full PDF of the book available so it can check anything that seems like an OCR error against the original. We've agreed on some conventions (using the YIVO transliteration for Yiddish names, or Russian where appropriate; putting short explanatory notes in brackets; etc). We maintain a list of all person and place names for reference so that we don't end up with different variations in different places.
I also give the same information to ChatGPT 5.6-Luna, and read the two in parallel. They generally agree very well, and when they differ it has been helpful to start digging into specific questions ("what Yiddish word is used here and how could it be translated?" "Do you think the author means xxx or yyy?"). Occasionally I just like ChatGPT's wording better and substitute it, or blend the two, but mostly I've been happy enough with what Claude generates. I adjust the formatting for the page numbers, section breaks, headings, etc.
Then I look for anything that could use more explanation, and add side-notes. I try to identify places, famous people, and helpful context, with links where useful.
I'm also having Claude update two auxilliary "artifacts" with each new chapter. One is a map, which I've found very helpful to get a sense of how all the unfamiliar places names relate to each other. From time to time I've asked for revisions to how it looks and interacts, eventually I'll come up with a version that can be connected with the online and print versions of the book.
The other artifact is a family tree / cast of characters. I have let Claude do its own thing with this so far, but eventually will think about how to make it into a useful reference to include with the book.