DEV Community

Cover image for EcoTwin: An AI Climate Coach for Real-World Emissions Cuts
jaysid97
jaysid97

Posted on

EcoTwin: An AI Climate Coach for Real-World Emissions Cuts

DEV Weekend Challenge: Earth Day

What I Built
Most climate tools diagnose. EcoTwin prescribes.

EcoTwin is a personalized climate action coach that turns a few everyday inputs, like commute habits, food choices, home energy use, and travel frequency, into a practical action plan with estimated annual CO2e savings.

Instead of stopping at a “guilt score,” EcoTwin estimates a baseline footprint, shows a before/after projection, and recommends the highest-impact changes first.

Users get:

  • A baseline annual footprint estimate
  • A personalized set of top climate actions
  • Projected annual CO2e savings
  • A concise AI coaching summary powered by Gemini (with fallback if a key is unavailable)

Demo
Live demo:https://ecotwin-qsgv.onrender.com/

Quick walkthrough:

  1. Open the app
  2. Enter a city and lifestyle inputs
  3. Click Generate My Climate Plan
  4. Review before/after footprint and recommended actions
  5. Read the AI coach summary

Code
Source code is in the project folder and includes:
-Backend:Flask API for scoring, recommendations, and AI summary
-Frontend:HTML/CSS/JS dashboard with interactive results
-Data:Curated action library with estimated CO2e savings

Core implementation highlights:

  • Transparent footprint estimation model
  • Rule-based personalization by user profile
  • Gemini API integration for natural-language coaching
  • Stable local fallback summary for reliability

How I Built It
I built EcoTwin as a focused full-stack Flask app to keep it easy to run, easy to demo, and easy to judge.

Architecture flow:

  1. Frontend collects a lightweight user profile
  2. Backend computes annual baseline emissions
  3. Action library is filtered by relevance and sorted by impact
  4. App calculates projected reductions and renders before/after metrics
  5. Gemini generates a short personalized coaching summary

Design decisions:

  • Kept input friction low so users can get value in seconds
  • Prioritized practical behavior-change suggestions over abstract climate theory
  • Added clear before/after visuals to make impact tangible during a live demo

Prize Categories
This submission is for:

  • Best Use of Google Gemini
  • Best Use of GitHub Copilot

Team
Solo submission.

Top comments (5)

Collapse
 
elisxu profile image
Elis

Amazing🙌🏻 it would be cool if this would be sth we could use, but what would it be benefit for the users in their daily lives? If I may

Collapse
 
jaysid97 profile image
jaysid97

sure

Collapse
 
elisxu profile image
Elis

Could you please explain??

Collapse
 
theeagle profile image
Victor Okefie

The "climate twin" framing is the right hook. Most climate tools give you a guilt score — here's how much you're damaging the planet. You gave a projection and a path. That's the difference between diagnosis and prescription. The low input friction matters because the barrier to action isn't knowledge — it's the feeling that individual changes don't add up. The before/after visual makes the math visible. That's not a dashboard. It's a mirror with a plan.

Collapse
 
jaysid97 profile image
jaysid97

Appreciate this. That diagnosis vs prescription distinction is exactly what I was aiming for. I wanted EcoTwin to feel less like a guilt meter and more like a practical weekly action plan, with visible before/after impact so small changes feel real and cumulative.