EcoQuest

EcoQuest turns real-world environmental action into a competitive game — complete eco-tasks, submit photo proof, earn points verified by AI, and compete for weekly cash prizes.

Comment

Inspiration

We've all seen the problems rising CO2 levels, plastic-choked oceans, disappearing forests. But for most people, especially students, the gap between caring about the environment and actually doing something feels enormous. We wanted to close that gap. The idea behind EcoQuest was simple: what if environmental action felt as engaging as a game? What if every beach cleanup, every tree planted, every recycling run earned you something real — points, recognition, even cash? We were inspired by the way fitness apps like Strava turn exercise into a social competition, and asked ourselves why nothing like that existed for eco-action. EcoQuest is our answer.

What it does

EcoQuest is a gamified environmental action web platform where users complete real-world eco-tasks such as beach cleanups, tree planting, recycling, and more — and upload photo or video proof to earn points. An AI verification system powered by the Gemini API KEY analyzes each submission, matches it against the task description, checks for fraud, such as duplicate or stock images, and assigns a confidence score. Scores above 80% are auto-approved; 60–79% are flagged for admin review; below 60% are rejected, with one appeal allowed.

Users browse tasks organized by environmental category (Water, Air, Land, Wildlife) and difficulty tier (Bronze, Silver, Gold). Completing tasks builds streaks that multiply points, and long-term progress unlocks a level system — Eco Newcomer, Eco Warrior, and Earth Champion. A live leaderboard resets weekly, with the top three players winning real cash prizes. A personal impact dashboard translates points into real environmental metrics: trees planted, plastic diverted, coastline cleaned, and CO2 offset. A global heatmap aggregates every user's contributions into one live counter visible on the homepage.

How we built it

We designed EcoQuest as a full-stack web application. The frontend is built with React.js and Tailwind CSS, following a nature-inspired design system with a card-based UI, custom difficulty tier badges, and a persistent navigation bar across all five core pages. The backend runs on Node.js and Express, handling the REST API, authentication logic, and business rules like streak calculations and weekly leaderboard resets.

We used PostgreSQL via Supabase for our relational database, storing users, tasks, submissions, and point histories. Media uploads — photos and videos submitted as proof — are stored on AWS S3 or Cloudflare R2. Real-time features like the live leaderboard countdown and activity feed are powered by Socket.io WebSockets. The geographic heatmap on the homepage and admin dashboard is built with Leaflet.js. The AI verification engine calls the Anthropic Claude Vision API, passing each uploaded image alongside the task description and returning a confidence score that drives the approval logic. Authentication is handled via Firebase Auth, supporting Google and GitHub OAuth. The frontend is deployed on Vercel and the backend on Railway.

Challenges we ran into

The hardest problem was designing the AI verification system to be fair. A simple pass/fail system felt too harsh — lighting conditions, camera angles, and image quality vary enormously in real-world submissions. We landed on the confidence threshold model (80% / 60% / below 60%) so that borderline cases go to a human admin rather than being auto-rejected, which felt more just and kept users from feeling punished for imperfect photos.

Fraud detection was another challenge. We needed to prevent users from submitting the same photo repeatedly or using stock images pulled from the internet. We addressed this with perceptual hashing (pHash), which detects visually similar images even when they've been slightly cropped or recolored. Balancing the weekly reset leaderboard with long-term progression was also tricky. We wanted weekly competition to feel urgent, but we didn't want users who fell behind to feel like their efforts disappeared. The solution was separating weekly points (which reset) from total points and level badges (which persist), so every user always has a reason to come back.

Accomplishments that we're proud of

We're proud that EcoQuest isn't just a concept — it's a fully specified, production-ready platform with real technical depth. The AI verification pipeline is the piece we're most excited about: using Claude Vision to actually analyze whether a photo matches a task description is genuinely novel in the environmental space. We're also proud of the impact conversion engine, which translates raw point data into meaningful environmental metrics using real-world science data — so users can see "I've diverted 3kg of plastic from landfills" rather than just "I have 600 points." That translation from abstract numbers to real-world impact is what makes EcoQuest feel meaningful rather than just gamified.

What we learned

Building EcoQuest taught us how much UX design matters for behavioral change. It's not enough to build a platform that works — it has to make taking action feel good in the moment. Every design decision, from the streak multiplier to the podium layout on the leaderboard, was made to answer the question: does this make a user more likely to go pick up trash tomorrow? We also learned a great deal about AI confidence calibration — understanding that a model's output is a probability, not a verdict, and designing systems that handle uncertainty gracefully rather than hiding it. And working within real environmental science data for the impact conversion rates gave us a new appreciation for how hard it is to measure ecological impact accurately.

What's next for EcoQuest

The immediate next step is a working MVP — taking this design spec and shipping a live, testable version. Beyond that, we want to explore partnerships with environmental nonprofits that could sponsor task campaigns or verify impact claims, giving EcoQuest institutional credibility. We'd love to add school and organization accounts, so teachers can run EcoQuest challenges with their classes. On the technical side, we want to improve the AI verification model with fine-tuning on environmental task data, and expand the heatmap into a full environmental impact report by region. Long-term, the vision is a platform where millions of small individual actions add up to measurable, documented, real-world environmental change.

Built With

Share this project:

Updates