This is a submission for Weekend Challenge: Passion Edition
What I Built
Spark is an AI-powered reflection experience that helps people discover recurring patterns in what naturally gives them energy, curiosity, and meaning.
Most "find your passion" quizzes ask a handful of multiple-choice questions and return a flattering personality label. I wanted to build something different.
Spark guides the user through a short interview covering four dimensions:
- ⏰ Time
- ⚡ Energy
- 🧠 Curiosity
- 🤝 Purpose
Instead of treating passion as something hidden waiting to be discovered, Spark treats it as a pattern already present in a person's choices.
After the interview, Gemini analyzes the responses, looks for recurring themes, weighs evidence across answers, considers alternative explanations, and generates a structured report including:
- Spark title
- Confidence score
- Spark DNA
- Supporting evidence
- Counter-evidence
- Blind spots
- Suggested experiments
- Final conclusion
The report intentionally presents its conclusions as hypotheses rather than absolute truths, encouraging users to validate them through experience.
Demo
Live Demo
Code
GitHub Repository
shravzzv
/
spark
Spark helps you discover what consistently gives you energy through guided reflection and AI-powered pattern recognition. Instead of telling you what your passion is, it reveals the patterns hidden in your own experiences.
⚡ Spark
Passion isn't a lightning bolt. It's a pattern.
Spark is an AI-powered reflection experience that helps people discover recurring patterns in what naturally gives them energy, curiosity, and meaning.
Rather than asking "What's your passion?", Spark asks a better question:
What do your choices repeatedly reveal about you?
Through a short guided interview and structured reasoning powered by Gemini, Spark generates an evidence-based report describing the user's strongest motivational patterns, supporting evidence, possible blind spots, and practical experiments to validate its conclusions.
Why Spark?
Many passion quizzes feel arbitrary.
They often:
- ask superficial multiple-choice questions,
- produce flattering personality labels,
- provide little evidence,
- and present conclusions as certainty.
Spark takes a different approach.
It treats passion as an observable pattern rather than a hidden trait.
Instead of attempting to "detect" a person's calling, Spark analyzes recurring themes across their own reflections.
Every conclusion is grounded in the user's…
How I Built It
Spark is built with:
- Next.js App Router
- TypeScript
- React Hook Form
- Zod
- Motion
- shadcn/ui
- Google Gemini
The application uses a multi-stage interview instead of one large form. Each stage focuses on a different aspect of motivation and stores its data in a single typed React Hook Form instance.
Once completed, the questionnaire is transformed into a readable interview transcript before being sent to Gemini.
Rather than asking Gemini to simply summarize the answers, I designed the prompt to reason about the user's responses. It is instructed to:
- identify recurring patterns,
- weigh evidence,
- explain its reasoning,
- consider competing explanations,
- avoid flattery,
- acknowledge uncertainty,
- generate practical experiments.
Gemini returns structured JSON matching a predefined schema rather than free-form text. The response is validated before being rendered into the final report.
This approach allowed the UI to remain completely data-driven while ensuring reliable outputs.
One of my favorite implementation details is reconstructing the original interview before sending it to Gemini. Instead of receiving anonymous JSON fields, the model sees each original question followed immediately by the user's answer, providing much richer context for its analysis.
Prize Categories
🏆 Best Use of Google AI
Gemini is the reasoning engine behind Spark.
Rather than acting as a chatbot or text generator, Gemini performs structured analysis over the user's reflections. It identifies recurring motivational patterns, evaluates supporting and conflicting evidence, estimates confidence, proposes experiments, and returns validated structured data that powers the entire report.
The goal wasn't simply to generate text—it was to build an experience that feels thoughtful, evidence-based, and genuinely useful.
Building Spark reminded me that passion probably isn't something we stumble upon one day.
It's something we can gradually notice by paying attention to the patterns that already exist in our lives.
I'd love to hear your thoughts and feedback!
Top comments (0)