Overview
This project began with a report inspired by Nicholas Felton's Personal Annual Reports and the quantified-self movement. I wanted to turn the traces of a year into something worth reading and keeping, not another activity dashboard.
The finished system combines Apple HealthKit, Spotify, photos and their metadata, Strava, GitHub, calendars, AI-platform activity, and other personal records. It keeps the source data local and produces findings that retain their coverage and provenance.
The Grounding Rule
The project uses one rule to separate useful AI assistance from plausible invention: the model may write the query; it may not write the answer.
AI can propose a comparison, trend, correlation, or grouping. Every quantity and claim must come from an operation executed against the records. Coverage travels with the result, because a correct calculation over weak evidence is still a weak finding.
Beyond One Report
Building the report produced a reusable personal data platform. It collects source-faithful observations, converts them into common forms, aligns them on a daily timeline, and lets different consumers ask grounded questions of the same archive.
The Japan travel log proved that separation mattered. It reused photographs, locations, dates, and steps from the annual-report platform to tell a different story without rebuilding the data pipeline.
Read the Project Story
I wrote about why this project stayed on my backlog for years, how AI changed the cost of building it, and the safeguards needed to keep the report honest in Building the Annual Report I Had Put Off for Years.