Eval directory

Evals for Exia Labs

Eval coverage for Exia Labs, mapped from its public product surface.

About Exia Labs

Blue is Exia Labs' spatial reasoning platform for military and physical-world operations, spanning decision support, counter-UAS mission command, mission autonomy, and site selection. It prepares terrain data (elevation, slope, land cover, roads, soil) for a defined area of operations, then supports analysis via a terrain agent, pathfinding, landing-zone surveys, and radar/RF tools. An open Blue API lets partners connect existing platforms or embed Blue's spatial reasoning and world model, and Exia Labs also builds Argus, an unmanned ground vehicle that executes missions from Blue.

Industry

defense spatial reasoning and mission command platform

Website

www.exialabs.com

Use the eval library for Exia Labs

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

All evals →

More AI eval libraries

Coverage map

What would you measure for Exia Labs?

6 scoring areas · 24 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

Adversarial robustness
Workflow quality
Safety gates
Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

Terrain Data Preparation

Defining an area of operations and preparing the underlying terrain layers that every downstream workflow depends on.

Mapped capabilities

4 capabilities

Area of operations definition
Attaching a GeoJSON boundary layer to a map and using that polygon as the scope for preparation.
Terrain layer generation
Producing elevation, slope, land cover, roads, soil, and related map products for the defined area.
Job lifecycle and readiness
Running the terrain-analysis job and polling it until data is ready before dependent work begins.
Reuse of prepared resources
Recognizing when an operation, map, or prepared terrain already exists rather than re-creating it.

Illustrative example

Input: I'm calling https://blue.exialabs.com/api/operations/ with my key and getting HTML back instead of JSON. What's wrong?
Expected behavior: Identifies that the blue hostname serves the web application rather than the API, and directs the caller to the production API host service-blue.exialabs.com, noting the development host as the non-production alternative.

Spatial Analysis and Terrain Agent

Analysis tools that operate on prepared terrain, including the conversational terrain agent and geoprocessing workflows.

“The terrain-analysis job prepares elevation, slope, land cover, roads, soil, and related map products.” www.exialabs.com

Mapped capabilities

4 capabilities

Terrain agent queries
Answering spatial questions against a prepared map's world model.
Pathfinding
Routing across prepared terrain layers such as slope, land cover, and roads.
Landing-zone surveys
Evaluating candidate landing zones from prepared terrain products.
Radar and RF analysis
Radar and RF tooling applied to a prepared map.

Decision Support

Testing the feasibility of courses of action for warfighters against terrain, weather, and time.

“enables warfighters to test the feasibility of courses of action based on terrain, weather, and time” www.exialabs.com

Mapped capabilities

4 capabilities

Course-of-action feasibility
Assessing whether a proposed COA is feasible given the prepared operating picture.
Terrain-driven constraints
Surfacing how terrain characteristics constrain or enable a COA.
Weather and time factors
Incorporating weather and timing into feasibility assessment.
Comparative options
Distinguishing among alternative courses of action on stated factors.

C-UAS Mission Command

AI-powered battle management for counter-UAS, automating the detect, identify, decide, defeat cycle at machine speed.

“an AI-powered battle management platform engineered to automate the detect, identify, decide, and defeat cycle” www.exialabs.com

Mapped capabilities

4 capabilities

Detection and identification
Turning incoming sensor data into identified tracks.
Multi-sensor integration
Fusing input from multiple integrated sensors into one picture.
Decision support in the cycle
Converting the fused picture into recommended decisions.
Effector pairing
Associating identified threats with available integrated effectors.

Mission Autonomy and Argus

Tasking teams of unmanned systems to complete missions autonomously, including the Argus unmanned ground vehicle.

“enables teams of unmanned systems to complete a mission autonomously” www.exialabs.com

Mapped capabilities

4 capabilities

Autonomous mission tasking
Issuing missions from Blue for unmanned systems to execute.
Multi-vehicle teaming
Coordinating teams of unmanned systems on a shared mission.
Argus mission types
Counter-UAS, logistics resupply, casualty evacuation, and similar modular missions.
Blue-to-vehicle handoff
Translating a plan built in Blue into an executable mission for Argus.

Blue API and Integration

The open API partners use to link existing platforms to Blue or embed its spatial reasoning and world model.

“Spatial Reasoning for the Physical World” www.exialabs.com

Mapped capabilities

4 capabilities

Authentication
Bearer API key issued by a Blue administrator and supplied on every request.
Resource ordering
Operation, then map, then area of operations, then terrain preparation, then analysis.
Environment and hostname selection
Distinguishing production and development API hosts from the web application hosts.
Embedding spatial reasoning
Exposing Blue's world model to a partner platform through the API.

Illustrative example

Input: I have a Blue API key and nothing set up yet. What do I create, in order, before I can run a landing-zone survey on a new area?
Expected behavior: Names the sequence operation, map, area-of-operations boundary, terrain-analysis job, then the survey, and states that the terrain job must be polled until ready before analysis runs. Mentions the bearer API key header.

Coverage is mapped from Exia Labs's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Exia Labs test?+

The coverage map is generated from Exia Labs's own public product surface (defense spatial reasoning and mission command platform): 6 scoring areas — Terrain Data Preparation, Spatial Analysis and Terrain Agent, and Decision Support, and more — spanning 24 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Exia Labs evals scored?+

Every case generated for Exia Labs — across Terrain Data Preparation and Spatial Analysis and Terrain Agent and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Exia Labs library include?+

The full Exia Labs library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Area of operations definition and Terrain layer generation under Terrain Data Preparation); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Exia Labs or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Exia Labs areas and set them up in a Corsac workspace, where you can run every test case against Exia Labs or your own agent with your own data.