Inspiration
Plant problems are difficult to describe with a single photo. A damaged leaf may be caused by watering, pests, light, soil, or the surrounding environment, and anxious plant owners rarely know which details matter.
What it does
Jardinly Live is a smartphone-first AI plant doctor. It guides the user through a calm live scan, asks for useful views of the whole plant, leaves, stems, soil, and environment, and turns the collected evidence into a structured diagnosis and personalized rescue plan.
The experience combines real-time voice guidance, visual evidence capture, plant identification, diagnosis, care planning, progress tracking, reminders, and an AI plant-care conversation in one PWA.
How we built it
The product is built with React, TypeScript, Node.js, PostgreSQL, realtime multimodal model integrations, and Google Cloud Run. During OpenAI Build Week, GPT-5.6 and Codex were used to extend an existing proprietary Jardinly foundation with new product design, reliability improvements, model fallback behavior, test coverage, architecture documentation, and a production-ready mobile experience.
The pre-hackathon foundation and Build Week additions are documented separately in the repository and commit history.
Current status
Jardinly Live runs as a working mobile PWA. The live scan, evidence collection, diagnosis flow, rescue plan, garden, progress tracking, and AI chat are integrated into the deployed product. Additional Build Week features and the final demo narrative are still being completed before submission.
Built With
- azure-openai
- gemini-live-api
- google-cloud-run
- gpt-5.6
- node.js
- openai-codex
- postgresql
- react
- typescript
