Building a tag-based journaling system (instead of text or habits)
I’m currently building TagJournal, a different approach to journaling and self-tracking.
The core idea is simple:
most tools sit on two extremes:
free-form journaling → rich but time-consuming and hard to analyze
habit trackers → structured but too rigid and lossy
I’m trying to bridge that gap with a tag-based model.
Instead of writing entries or ticking habits, each day is represented as a set of tags (+ optional values).
For example:
mood: good
work: deep-focus
sleep: 7
workout
From a data perspective, it behaves more like a lightweight event log than a journal.
This unlocks a few interesting things:
low friction input (seconds per day)
flexible schema (users define their own rich-tags)
structured data for analysis without forcing rigid templates
On the backend side, I’m treating everything as time-series data:
each “tag usage” is a timestamped record
optional numeric values allow aggregation (averages, trends, correlations)
the goal is to make querying fast enough to power real-time insights
One challenge I’m focusing on is balancing:
flexibility (user-defined tags)
vs meaningful analytics (which usually require structure)
I’m also exploring:
how to surface correlations without overwhelming the user
how to keep the UX faster than both journaling and habit tracking
sync across devices with an offline-first approach
For users who prefer writing, each tag also supports long-form text with a distraction-free editor, plus AI features like natural language → structured tags, AI insights, and automatic day summaries.
Curious if anyone here has worked on similar “semi-structured personal data” problems, or has thoughts on modeling this kind of dataset.
