Skip to content
FR

Organization app

Goals

Personal project

View the app
Calendar view of the Goals app on a laptop.

Goals is a smart habit tracker that links daily actions to big life goals, powered by an AI assistant that finds hidden patterns in your progress. Most goal-tracking apps treat goals as a flat, disconnected to-do list: users tick boxes, but quickly lose sight of why they are doing it.

Brief and goal

Link daily actions to big life goals, rather than treating them as a flat, disconnected to-do list where you quickly lose sight of why you tick the boxes.

Role and responsibilities

Personal project: UX/UI design, full-stack development and AI integration.

Credits

Personal project. The UX/UI design, the full-stack development and the AI integration are my work.

Use of AI

I used Google Antigravity to program and debug the app, and the Gemini API, with a specialized prompt, to analyze daily check-ins in the app; I used Claude Code to add this sheet to the portfolio, from my case study. My personal contribution includes the UX/UI design, the data structure, the system prompt and the full-stack development.

Production steps

  1. UX/UI: a hierarchy of goals

    Instead of a flat list, the interface leads users to link small habits to parent goals. I used a clean, upscale dark-mode aesthetic to keep the interface focused and distraction-free, while visually emphasizing the hierarchy.

    Goals tree of the Goals app: main goals such as « Join the running club » and « Write a book », linked to their sub-goals.

  2. Development: flexible daily check-ins

    Built with React and Firebase, the app offers an evening reflection based on personalized questions. Over time, this creates a rich, structured dataset, without feeling like a chore.

    Evening check-in of the Goals app, step 2 of 7 « Today’s stats »: wake-up and bed times, hours of sleep, happiness and productivity levels, and exercise.

  3. AI system: contextual analysis

    The user’s check-in history is formatted and sent securely to a large language model API with a specialized prompt. The system looks for correlations (for example, « the days you sleep 8 hours are correlated with completing your coding habit ») and returns structured, actionable insights rather than generic advice.

    The « AI Insight Chat » panel of the Goals app: a personal data analyst suggests questions about sleep, productivity and happiness.

  4. Structured data for the AI

    For the analysis to give concrete insights rather than vague advice, check-ins combine free text with binary and numerical entries; the system prompt then enforces the « Observation → Correlation → Action » format.

    Calendar of the Goals app: a 67% goal completion rate, successful or missed days, and bedtimes and exercise overlaid on the days.

  5. Deployment

    The front end is built with Vite and the app runs in production on Google Cloud Run; authentication and the database are handled by Firebase.

    Goals app open on its calendar view, on a laptop on a wooden desk.

Technologies

Challenges

At first, the AI gave vague, horoscope-like advice that was impossible to put into practice, because the check-in data was not structured enough.

Solutions

I restructured the database to require binary and numerical entries on top of text, and refined the system prompt to enforce a strict « Observation → Correlation → Action » format.

Final result

A web app online: hierarchical goals, daily check-ins and trend analysis through the Gemini API, built with React and Firebase.