LGND (pronounced "legend") is a geospatial AI company on a mission to make Earth understandable — for people and AI alike. We harness geospatial embeddings and large Earth observation models trained on satellite and aerial imagery to turn billions of images into a living, queryable record of the planet.

The result is infrastructure that delivers planet-scale datasets faster and more affordably than traditional pixel-by-pixel approaches. You can explore Earth through natural language chats, interactive maps, and developer APIs.

Traditional image classification required hundreds or thousands of annotated examples and custom-trained models for every new task to become expert “recognizers”. That meant months of data labeling and expensive ML infrastructure just to answer a simple question like "how many solar panels are in Texas?"

Large Earth observation models like Clay change the economics entirely. They're trained on massive corpora of satellite imagery and can generalize to new tasks with just a handful of reference examples. LGND wraps this capability into production-ready APIs and no-code applications — so organizations can get answers in hours instead of months, at a fraction of the previous cost.

Yes! LGND's API is designed to work with your own imagery sources, not just the built-in Sentinel-2 or NAIP feeds. You can submit custom raster tiles — from commercial satellite providers, drone captures, airborne sensors, or proprietary archives — and LGND will generate embeddings using the Clay foundation model.

This is particularly valuable for teams with access to higher-resolution commercial imagery (e.g., Vantor, Planet, Airbus) or specialized sensors (SAR, LiDAR-derived rasters, hyperspectral) that require custom analysis beyond publicly available data.

LGND works with standard geospatial raster formats. Key considerations for bringing your own imagery:

Format — GeoTIFF and cloud-optimized GeoTIFF (COG) are the preferred formats

Tile size — typically, chips are processed at 256×256 pixels; your imagery will be tiled automatically at this dimension but other dimensions are possible to set

Bands — any multispectral imagery with wavelengths within or near the pre-training range can be used; the model's wavelength encoding ensures robust generalization

Projection — standard projections (WGS84, UTM) are supported

Contact the LGND team for guidance on integrating custom sensor types or non-standard imagery sources.

LGND generates, hosts, and indexes the resulting embeddings for you. Once processed, your embeddings are accessible via the API for vector similarity search, classification, and downstream analytics — without you needing to manage a vector database or embedding infrastructure.

Embeddings are versioned and linked to their source imagery metadata (location, timestamp, sensor), making it straightforward to track lineage and run time-series analyses across imagery updates.

Yes. A common workflow is to use LGND's hosted Sentinel-2 or NAIP embeddings for broad-area search and screening, then refine results using higher-resolution custom imagery for the subset of locations that require more detail. This hybrid approach balances coverage and resolution cost-effectively.

The LGND API allows you to work across multiple embedding sets within the same project, enabling multi-source workflows without duplicating annotation work.

LGND currently hosts the Clay foundation model. Additional open-source models will be available in the near future. 

LGND can be run on any sized area. The unit of analysis is a raster tile. A raster tile represents a single remotely sensed image (satellite, aerial, drone) for a specific location on Earth and acquired at a specific time.

Chips are typically 256x256 pixels. A Sentinel-2 chip  that is 256 – where each pixel is 10 meters - would therefore have a corresponding area of 2.5km^2. 

LGND unlocks significant accuracy with just a few reference examples. Accuracy depends on many factors: how much training data is provided, how distinct an object is relative to its surroundings, and how variable the object is over space and time. It is rare for a model to work perfectly out of the box. Like with other AI tools, LGND’s analytics are refined through user prompting and feedback. 

Typically, model training requires a few minutes to hours. Inference takes a few minutes to hours. Both depend on the number of labels used (training) and the area of interest (inference).

The Clay model was trained on 10 bands from Sentinel-2 imagery, 10 bands from Landsat imagery, and all four bands of NAIP.

Wavelengths are encoded in the model. It can therefore extrapolate to wavelengths that are within or near the ranges used for pretraining.

You can run a model as many times as you’d like. If you're studying a phenomenon that changes frequently, you can run your model on each update of imagery. Landsat and Sentinel-2 offer updates roughly every five days. NAIP imagery updates every other year.

Models can be run each time new (cloud-free!) imagery becomes available. 

The LGND API gives developers programmatic access to the full embedding stack. The API is designed for integration into existing geospatial pipelines, AI applications, or data products. Visit the Documentation page for more technical details. Key capabilities include:

Embedding generation — process satellite tiles (LGND-hosted or your own) into vector representations

Vector similarity search — find tiles that match a query image or text description

Model training & versioning — launch supervised training jobs on curated label datasets

Inference — run trained models over any AOI and retrieve structured detection outputs

Human-in-the-loop — validate predictions and feed approved results back into training

LGND Discover lets you search Earth imagery using natural language. Instead of browsing maps manually, you describe what you’re looking for and Discover scans large-scale satellite imagery to surface relevant locations in seconds. It’s powered by LGND’s geofoundational intelligence, which understands both geographic context and visual patterns in imagery.

You type a plain-language query into the search bar describing what you’re looking for (a feature or change). You can include where you want to look (a place or region) and / or when you're interested in (recent changes). Some examples include “Find new solar arrays in Texas”, “Find parks with playgrounds in Denver", or “Find homes with tennis courts in Los Angeles”.

Discover automatically retrieves the geographic boundary (for example, the polygon for “Texas”) and searches imagery that intersects both your location and feature of interest

Discover previews 4 matches in the chat and displays up to 100 results on the map and in the grid. Each result is color-coded by confidence. Confidence reflects how closely the imagery matches your query based on embedding similarity. 

Yes. You can switch to a grid view to see image thumbnails for all results. Selecting a result shows: Image date, Data collection source, Similarity score.

You can also locate each result on the map and use a date slider to view how that location changed across acquisition years.

Discover supports three refinement methods:

1. Direct feedback

Use thumbs up/down on results. Re-running the search incorporates your feedback to improve relevance.

2. Follow-up prompts

Add constraints like “only in desert regions” or select suggested prompts to narrow intent.

3. Manual annotation

Draw examples directly on the map using the annotation tool. You can even label before/after imagery to specify a change pattern.

These signals are used immediately to improve subsequent searches.

If your query implies change (such as “new” solar arrays), Discover analyzes imagery across multiple timestamps and looks for transitions, like a feature going from absent to present. The reasoning behind these results can be inspected directly in the app through model reasoning traces.