CruzDex

🦎 turn your backyard into a Pokédex and your observations into real science.

Biodiversity is changing faster than we can document it. Species are shifting ranges, emerging earlier, disappearing quietly. Scientists know this is happening but don't have enough eyes on the ground to track it in real time.

The Solution

Cruzdex is a location-aware citizen science game where an AI agent, powered by NVIDIA Nemotron. Continuously analyzes real ecological data and generates personalized field challenges that send players to the exact right place, at the exact right time, to find something scientifically meaningful.

How It Works

Every morning, the Nemotron Agent, CruzDex's gamemaster, wakes up and reasons over:

• iNaturalist's entire observation history for your region
• Current and forecast weather
• seasonal phenology calendars
• live population abundance trends
• your personal Cruzdex gaps

Players go out, find the species, photograph it, and submit through the app, which automatically syncs the observation to iNaturalist, contributing it to the global biodiversity record.

The game awards XP. Science gets data. Everyone wins.

The Nemotron Agents

Nemotron doesn't just generate quests. It generates hypotheses and players are the instruments that test them.

This is the Nemotron showcase. Nemotron is a multi-agent system where specialized agents each own a distinct reasoning task:

• Phenology Agent — reasons about what should be appearing this week based on historical first-sighting dates
• Climate Agent — flags anomalies where current conditions deviate from 30-year normals, triggering early detection challenges
• Abundance Agent — tracks year-over-year population trends, generating documentation missions for declining species
• Boundary Agent — computes species range edges from historical GPS clusters, sends players to the frontier where ranges are shifting
• Gap Agent — knows your personal collection, filters challenges to species you haven't found yet
• Nemotron Orchestrator stage — synthesizes all five signals into ranked, personalized, scientifically justified challenges with geographic boundaries, timing windows, and XP rewards scaled to scientific value

Built With

• 19
• dom
• flask
• html
• inat
• javascript
• nemotron
• node.js
• python
• react
• supabase
• tailwindcss
• vercel
• vite