AIvot

What it does AIvot lets users quickly log realtime life updates like stress, mood, energy, and context tags. Those check-ins are then used to track mental wellbeing trends, detect distress signals, and improve the quality of the app’s insights and recommendations. It also gives users a clear daily pulse so they can see their own patterns instead of only reacting after things get worse.

How we built it We built it as a local-first experience in the frontend with persistent storage, a dedicated LifePartner panel, quick-prompt templates, a daily goal/streak view, and a timeline of recent updates. The scoring pipeline was wired to include the realtime LifePartner context, and we added clearer AI-vs-fallback pipeline state so it’s obvious when the app is using live AI or local heuristics. On the backend, we hardened the score parsing so the system is more resilient when model output is messy or rate-limited.

Challenges we ran into The biggest issue was that the AI pipeline did not always behave like a true live system. When Gemini quota limits kicked in, the score endpoint would fail and the app would silently fall back, which made the output feel static. We also found that the fallback logic was not using the new LifePartner signals strongly enough, so mental score changes and insights did not feel connected to user behavior.

Accomplishments that we’re proud of We turned LifePartner from a logging form into something more habit-forming and human. The app now gives users one-tap prompts, a visible daily goal, a streak, and immediate feedback that makes check-ins feel lightweight instead of like work. We also made the pipeline more transparent and made sure realtime life signals actually affect the score and insight flow, even when AI is unavailable.

What we learned We learned that people will only use a wellbeing feature willingly if it feels fast, forgiving, and rewarding in the moment. We also learned that mental health software needs graceful degradation: if the AI layer is unavailable, the product still has to be useful, explain what is happening, and stay emotionally coherent. Clear feedback and low-friction input matter as much as model quality.

What’s next for AIvot Next, AIvot should move from local-first intelligence to stronger backend persistence and smarter trend analysis. The most useful next step is to keep improving how LifePartner turns short daily check-ins into better predictions, clearer summaries, and more personalized support. After that, the product can expand into stronger reminders, deeper weekly reviews, and optional multi-device sync.

Built With

• ai
• and
• api
• auth
• backend;
• browser
• built-with-react
• css
• current
• database
• dedicated
• express.js
• for
• framer
• frontend;
• gemini
• google
• in
• is
• layer
• localstorage
• lucide
• motion
• no
• node.js
• on
• or
• persistence.
• polish;
• react
• scoring/chat;
• tailwind
• the
• typescript
• ui
• used
• vite