I Asked AI to Show Me My Life in 2050 — It Was Terrifying

This is a submission for Weekend Challenge: Earth Day Edition

I asked AI to show me my life in 2050.

It generated 3 versions.

One of them was… uncomfortable to read.

Try This (Takes 2 Minutes)

👉 https://tachyon-five.vercel.app

Then come back.

This will hit differently.

⚡ Quick Summary

• Answer 8 lifestyle questions → get 3 personalised futures
  


• AI generates messages from yourself in 2050
  


• You commit to one change before seeing results
  


• Focused on behaviour change through storytelling
  


• Built with Next.js, Gemini, Auth0, Supabase
  

🌍 What I Built

TACHYON is an AI-powered experience that shows you three possible versions of your life in 2050, based on how you live today.

Instead of showing climate data, it generates personal narratives from your future self.

The goal was simple:

Make climate change feel personal, immediate, and impossible to ignore.

🔮 The Three Futures

🔴 You Changed Nothing

• Your life continues unchanged
  


• Your city may no longer be livable
  


• Message appears corrupted
  

🟡 You Made One Change

• That one commitment matters
  


• Life improves, but remains fragile
  

🟢 You Went All In

• You changed more than expected
  


• Your environment remains intact
  

The Moment That Hit Me

“We had to leave the city you loved.”

That line stayed with me.

Because it felt real.

🤯 Why This Works

This project is built on one idea:

People don’t change behaviour because of data.


They change when the future feels personal.

🧠 Core Concept: Future Self Feedback Loop

1. Capture current lifestyle
   


2. Generate multiple futures
   


3. Let users emotionally experience outcomes
   


4. Create instant comparison
   

This compresses decades into minutes.

Demo

👉 https://tachyon-five.vercel.app

Code

👉 https://github.com/sushilkulkarni1389/tachyon

🏗️ How I Built It

Stack

• Next.js 14 (App Router + TypeScript)
  


• Gemini 2.5 Flash/ Gemini 3.1(fallback)
  


• Auth0 (secure access to AI APIs)
  


• Supabase (Postgres + JSONB storage)
  


• Tailwind + custom terminal UI
  


• Vercel deployment
  

Key Technical Decisions

1. Single AI Call for All Futures

Instead of multiple requests:

POST /api/transmit

Returns all three timelines in one structured response.

This ensures:

• Consistency


• Performance


• Better narrative cohesion

2. Prompt Engineering Insight

Initial mistake:

• Overly theatrical system prompts

Result:

• AI produced unreadable output

Fix:

• Neutral system role


• Strong few-shot example


• Context moved to user input

3. Commitment Before Outcome

The system asks:

“What will you actually change?”

Before showing results.

This creates:

• Emotional investment
  


• Ownership
  


• Stronger impact
  

4. Meaningful UI Design

• Glitch effects represent unstable futures
  


• Clean output represents stable futures
  

UI communicates state — not just visuals.

Prize Categories

This project is submitted for:

• 
  

  Best Use of Google Gemini

  
  
  
  
  
  • Generates deeply personalised, multi-path future narratives in a single request
    
  
  
  
  
  
  



• 
  
  

  Best Use of Auth0 for Agents

  
  
  
  
  
  • Secures AI endpoints and ensures controlled access to generation APIs
    
  
  
  
  
  
  

🧾 Conclusion

Most apps inform.

This one makes you feel.

And that’s what makes it different.

💬 What Do You Think?

Would something like this actually change behaviour?

Or is it just a powerful moment?

Curious to hear your thoughts 👇