Inspiration We were frustrated with to-do apps that just list tasks without helping you actually plan them — we wanted an AI that thinks alongside you, not just stores your data.
What it does It's a Gemini 3 Pro-powered productivity app that breaks tasks into subtasks, schedules your day based on energy patterns, verifies task completion with AI vision, and sends smart reminders — all through a conversational AI assistant.
How we built it React + TypeScript frontend with shadcn/ui, Supabase for auth/database/edge functions, Capacitor for mobile, and Google Gemini 3 Pro via Lovable AI gateway for all intelligence — task breakdown, scheduling, recommendations, and photo verification.
Challenges we ran into Getting the AI to return consistently structured subtask suggestions with realistic time estimates required careful prompt engineering and tool-calling schemas, and wiring real-time conversational context through edge functions was tricky.
Accomplishments that we're proud of The AI task assistant feels genuinely useful — you type "going to the vet" and instantly get actionable steps with timelines, and the photo proof verification using Gemini 3's vision adds real accountability.
What we learned Gemini 3 Pro's function calling is incredibly reliable for structured outputs, and building AI features as conversational flows rather than one-shot requests makes the experience far more natural.
What's next for AI Productivity Companion Collaborative task planning for teams, deeper calendar integration with Google/Outlook, and using Gemini's long context window to learn from months of user patterns for truly personalized productivity coaching.
Built With
- antigravity
- gemini
- javascript
- node.js
- react
- supabase

Log in or sign up for Devpost to join the conversation.