Back in 2021 I started a quiet personal project. A History of the Parish of Neilston is a local history book published in 1910 by David Pride, the kind of thing that matters enormously to the people it’s about and barely exists anywhere else. A few physical copies survive. Someone had scanned it and put it on archive.org, which was a start, but a scan isn’t really accessible. I wanted to produce a proper digital version with clean markdown, decent LaTeX formatting, usable images where possible.
I started working through it manually. Reading the scan, retyping, structuring chapters, cleaning up images. It was painstaking and slow, but I actually enjoyed it. There’s something satisfying about handling old text carefully. My focus fell away after six of the roughly nineteen chapters and it was left half-finished.
That was 2021, pre-AI, and literally a different game.
Fast forward to 2026 and I picked it back up. The OCR output from the scan was rough, technically readable but full of artefacts, broken lines, and formatting that needed a lot of work. What changed is that Claude and Codex turned out to be genuinely good at taking that messy output and formatting it into my existing structure. I kept the original grammar and language throughout, which matters for a document like this, but the mechanical work of structuring, tidying, and making it consistent is no longer a slog. I still handle the images manually, and some tables need hands-on attention, but the bulk of the text is moving at a pace I couldn’t have managed alone.
It’s not done yet, but it’s moving again. I’m not sure it would have been otherwise.
It’s a small thing in the grand scheme. Not a big open source project or anything commercially interesting. But a local history book from 1910 being properly digitised and readable feels worth doing.