I had wanted this for years: a way to turn daily data into a written record automatically. Nearly a decade ago, I was already building small tools around the GitHub Events API to generate monthly summaries of what I had done.
They worked, but they never lasted. Every version still depended on manual updates, and eventually I would stop. I kept running into the same wall: building something is easy compared with building something that keeps working with your real life.
Then I built a private LLM-powered prototype, and it genuinely moved me. For the first time, the output felt alive rather than mechanical. With my background in ETL and data integration, the architecture for deariary clicked almost immediately.
The space is still early, but the timing is right. LLMs make iteration fast enough to ship and improve in real time. After years as a full-stack engineer in startups, I wanted to turn what I know into something genuinely useful. deariary is that attempt: a diary that exists without asking you to become a disciplined diarist.
Founder, deariary