Born from a
Single Box of Letters.

The Spark

"My dad is suffering with early dementia. My mom wanted me to check out his decades of documents..."

It started with a simple request. A lifetime of memories—letters, diaries, notes—sat in boxes, gathering dust. I snapped a few dozen pictures with my phone, intending to read them later.

I brought them home. Three months passed. I hadn't read a single page.

I felt guilty. But the reality is, manual transcription is punishing. Squinting at handwriting on a phone screen, trying to decipher faded ink... the friction was too high. The memories were there, but they were locked away.

And we soon realized this wasn't just a personal problem. When we began working with the massive Thistlewood collection containing decades of sprawling documents, we found the sheer volume insurmountable for traditional historians as well.

The Breakdown & The Build

That's when our engineering team stepped in. I couldn't change my dad's condition, and we couldn't magically transcribe thousands of pages overnight for the Thistlewood researchers, but we could change the paradigm of archival access entirely. It started with endless late nights, drinking too much coffee, and writing scripts to try and automate what we physically couldn't do by hand.

• I built a custom text-extraction engine using advanced vision models to read the handwriting.
• I ran the text through a summarizer to understand the core themes.
• And then it hit me. Why just summarize? Why not tell the story?

When I ran those first few pages through the pipeline, I didn't get back a sterile transcription. The model had stitched the scattered fragments together, restoring the flow of his original timeline. For the first time, I could sit down and just read his early years. The memories were finally accessible.

That was the birth of Theirloom.