Explore the planet with an AI that speaks Earth Engine
Google Earth Engine is the world's most powerful platform for planetary-scale geospatial analysis. geeni is an AI expert that helps you unlock it—whether you're a researcher, developer, student, or just curious about what's happening on Earth.
What is Google Earth Engine?
Earth Engine combines a multi-petabyte catalog of satellite imagery and geospatial datasets with planetary-scale analysis capabilities. It lets you analyze how the Earth has changed over 40+ years —deforestation, urbanization, water quality, climate patterns—all from your browser.
Satellite imagery
Access decades of Landsat, Sentinel, MODIS, and hundreds more datasets. See how any place on Earth has changed over time.
Planetary-scale analysis
Run computations across entire continents without downloading a single file. Google's infrastructure handles the heavy lifting.
Environmental monitoring
Track deforestation, measure crop health, monitor water bodies, and detect changes in land cover across the globe.
Urban and climate studies
Analyze urban heat islands, map flood risk, model climate impacts, and support disaster response with real data.
JavaScript and Python APIs
Write analysis scripts in the Code Editor (JavaScript) or via the Python API. Publish interactive apps for anyone to use.
Free for research and education
Earth Engine is available at no cost for academic and research use. Commercial use is available through Google Cloud.
What can Earth Engine do?
Ask geeni about your city, your country, or any place you're curious about. Here are some things people explore every day.
See your city's green cover
Measure vegetation health using Sentinel-2 imagery. Compare seasons, track urban greening, or spot deforestation.
Try this →Map flood risk in your region
Combine SAR radar data with elevation models to identify flood extent and vulnerable areas after extreme weather events.
Try this →Track deforestation
Use the Hansen Global Forest Change dataset to quantify tree cover loss year by year and visualize the pattern.
Try this →Monitor air quality
Analyze MODIS aerosol products to understand air quality patterns and seasonal pollution cycles.
Try this →Classify land cover
Train a classifier on satellite imagery to automatically distinguish forests, cropland, water, and urban areas.
Try this →Measure urban growth
Compare nighttime lights or built-up area indices across decades to visualize urban sprawl.
Try this →Meet geeni
geeni is an AI assistant purpose-built for Earth Engine. It understands the APIs, knows the pitfalls, and draws from a curated knowledge base of documentation, forum answers, and best practices.
Under the hood, geeni uses retrieval-augmented generation (RAG) over 4,000+ curated knowledge chunks: API docs, developer guides, the EEFA textbook, and high-quality forum answers. Every response is validated against 145 test fixtures and checked for common anti-patterns like client-side computation traps and deprecated APIs.
How geeni learns and improves
geeni doesn't just answer questions—it gets better over time through a structured coaching loop. Every answer is evaluated, and patterns of failure become new training signal.
Answer
Generate a response using RAG and LLM reasoning
Evaluate
Score against a 6-dimension rubric with 145 test cases
Diagnose
Identify failure patterns and anti-patterns in traces
Patch
Generate targeted prompt patches that prevent recurrence
Tested against 145 real-world tasks
geeni is continuously evaluated against curated Earth Engine tasks spanning vegetation analysis, flood mapping, machine learning, time series, and more. Both Python and JavaScript code is validated by executing it against the real Earth Engine API.
99% criteria pass
102 of 103 golden test fixtures pass on the latest benchmark run, exceeding the 95% certification threshold.
145 test fixtures
103 curated golden fixtures plus 42 from the EEFA textbook, covering easy, medium, and hard tasks.
Cross-language validated
Reference code is tested in both Python and JavaScript to verify that Earth Engine computations produce identical results.
8 languages
The interface and test fixtures are available in English, Spanish, Chinese, French, Korean, Portuguese, Hindi, and Arabic.
Part of a larger agent ecosystem
geeni doesn't work alone. It integrates with Google's AI ecosystem and can hand off to specialized experts, maintaining conversational continuity throughout.
A natural language interface for agents
geeni is a discoverable tool. Other AI agents can invoke geeni's capabilities through a natural language interface—no special SDK required. Ask a question in plain English, get expert Earth Engine analysis back.
Ready to explore?
Whether you're writing your first Earth Engine script or debugging a complex analysis pipeline, geeni is here to help. Start a conversation and see what the planet looks like through data.