Notice what happened & what repeats
A journal for quick notes on how events made you feel. The journal offers a reading of each entry and notices what repeats over time. You correct it, and the final interpretation stays with you.
What happened?
A question to sit with
History
Each entry keeps what you wrote and the feelings you settled on. Open one to see what came after, then look back at it.
Patterns
Not “this is your pattern”. Just some things that may be worth noticing, drawn from the version of each entry you confirmed.
Possible patterns
Set aside by you
Across your entries
Each ring is one entry, the day’s first entry on the outside. Select a day to see it hour by hour.
Higher dots were more pleasant. Select one to open the entry.
Behind the design
Read about the inspiration, the design question and how the journal takes part without taking over.
Read about this projectAbout this project
Emotion Journal is a reflective journal for quickly documenting emotional reactions to events and discovering patterns across time.
How the journal takes part
- You capturea moment in your own words
- The journal interpretspossible emotion, context and trigger
- You correctevery interpretation remains editable
- The journal comparespossible patterns emerge across your history
- You reflectpatterns are hypotheses, not conclusions
The design question
The question here is not how AI can tell someone what they are feeling. It is how AI can help someone notice patterns in their own experience while preserving their authority over interpretation.
So the AI is framed as a hypothesis generator. It proposes; the person writing the journal decides. Its language is tentative on purpose: “I noticed…”, “This may be connected to…”, “One possible reading…”.
Three roles for the AI
- Interpreter
- Reads one entry and proposes a possible emotion, trigger, intensity and a short observation.
- Pattern Finder
- Looks across entries the person has confirmed and proposes recurring relationships, such as an emotion that keeps appearing next to the same kind of situation.
- Reflection Layer
- Turns observations into a question. It never tells someone what they should feel, what diagnosis they may have or what to do.
Human in the loop
Every reading can be confirmed, partly accepted or rejected, and every label stays editable. When someone changes a reading, their version is what gets stored, and it is their version the Pattern Finder learns from. Patterns can be marked meaningful, maybe, or set aside.
What is simulated here
This portfolio demo simulates the AI interaction. The readings come from transparent, rule-based matching against a small vocabulary of emotions and triggers, and the sample journal is fictional. No text you write is sent anywhere: entries are kept in your browser’s local storage.
Privacy, and the original prototype
Personal journal content should remain private whenever possible. The original prototype explored local-model processing, running language models on the writer’s own machine through Ollama and LM Studio, so private journal content would not need to be sent to a third-party AI service. That version is kept in the repository’s prototype folder.