AI-ASSISTED · PURPOSE EXPLORATION

IKIGAI

An AI-assisted web app that helps users explore possible Ikigai directions from two starting questions: what they love and what they are good at.

React · JavaScript · Vite · Gemini · Cloudflare Workers · SVG

Ikigai landing page with a colorful four-circle purpose diagram
Interactive Ikigai results diagram with four overlapping colored circles

PROJECT SNAPSHOT

Two questions become an explorable direction

Type
AI-assisted responsive web application
Status
Live independent Cloudflare deployment
Technologies
React, JavaScript, Vite, Gemini, Cloudflare Workers, SVG
  • Gemini-generated directions grounded in two user responses
  • User-selected path and structured final interpretation
  • Interactive four-circle SVG diagram
  • Browser-local session persistence and resume
  • PNG, PDF, and plain-text exports
  • Server-side credentials, validation, retries, and rate limiting
LIVE DEMO
Four Gemini-generated Ikigai paths with one selected Ikigai results page showing the interactive diagram and selected detail

Experience and AI grounding

Traditional Ikigai exercises ask people to define all four areas before they may know how to answer them. This experience begins with what the user loves and what they are good at, then offers several grounded possibilities for the user to choose rather than declaring a single purpose.

The selected direction becomes an interactive interpretation of love, ability, practical need, and potential paid value. The wording remains exploratory, and prompts instruct Gemini not to invent interests, experience, values, or personality traits.

Engineering details

Structured response schemas and runtime validation keep both AI stages predictable. Requests have bounded loading states, retry temporary model failures, and fall back to a second model when appropriate. Answers, paths, and results remain in local browser storage.

A native Cloudflare Worker keeps the Gemini credential server-side, serves same-origin API routes, applies request-size and rate limits, and delivers the Vite application with SPA fallback. AI output remains a starting point for reflection and depends on external model availability.