One of the most frustrating things about researching Salonika records at the National Archives is that they haven’t been digitised for easy online reading and text searching. It would make life so much easier if records – such as company war diaries – were digitised and readily available online.
As things stand, the only way to read the documents you need is to:
- Order them in advance.
- Make the trip to Kew.
- Photograph the pages yourself.
Admittedly, it’s not an unpleasant experience. The staff at Kew are unfailingly friendly, helpful, and efficient, and it’s a lovely place to spend a day. However, once you get those photos home, they aren’t always easy to read on a screen and searching the text for key terms is impossible. Your only alternative is the hard graft of spending long, tedious hours decoding the handwriting and typing it up manually. That is a real slog. But this is where artificial intelligence can, possibly, lend a hand.
Earlier this year, I travelled to Kew to view the war diaries of the 17th Field Company, Royal Engineers. I photographed each page on my phone -a task that took about an hour – leaving me with dozens of images of tricky handwriting that my computer couldn’t search.
Rather than transcribing everything by hand, I uploaded the images to Google Gemini and asked it to extract the text while preserving the tabulated format. What would normally have taken days took roughly two hours instead. While not the most thrilling task, it was a huge timesaver compared to manual entry.
On the whole, the results were impressive. In several cases, the AI managed to read handwriting I’d found completely impenetrable. Yes, it did make the occasional mistake and, like me, it struggled with some decoding. At a guess, I’d say it was about 95% accurate.


It’s not a complete solution as there’s still time needed for checking the transcription but, if you are facing a similarly laborious task, using AI to transcribe photographed records could be an effective and time-saving way to work.




