Spatial Analysis
Natural-language spatial analysis
Spatial analysis is Nexma's GIS capability, run in natural language. Ask a question about terrain, infrastructure, or demand and Jax becomes your analyst — querying, joining, and analyzing every layer of the physical world and returning the answer as a map instead of a spreadsheet. There is no query language to learn and no analyst bottleneck to wait on. The GeoEngine does the spatial work underneath; you only ask the question.
It is the same operating system — DataStore, Ontology, and Jax — with the GeoEngine handling the geometry, the joins, and the statistics.
What you can do
- Ask in language. Describe the question; Jax composes the SQL and spatial operations, runs them server-side, and explains the result.
- Query across every source. Joins span your uploads, the DataStore world model, live telemetry, and deep historical data in a single statement.
- Get answers as maps. Results land as live, styled map layers with automatic statistics — ready to act on inside your project, never a static export.
- Analyze access and reach. Isochrones, drive-time, and reachability along the real road network answer "how far can we serve from here?"
- Find patterns. Density, clustering, and hotspot detection surface where demand, risk, or activity concentrates across space and time.
Core concepts
Traditional GIS makes a human drive every step — wrangle projections, write the joins, stitch tools together — while the data sits in silos. Spatial analysis in Nexma collapses that into a question, because every layer already lives in or connects to one world model.
- The engine is the analyst. SQL and PostGIS-style geometry run underneath, but you do not touch them unless you want to. Jax interprets the question, picks the right methods, and returns a styled, sourced map.
- Typed when it can be. With an Ontology loaded, queries become semantic — by entity, property, and unit. Without one, analysis still works on raw uploads with no setup.
- Reproducible by default. Every result layer keeps its sources and its method, so an answer can be audited, shared, and re-run.
SQL is one of the things the GeoEngine runs underneath. The engine interprets the question, picks the spatial methods, queries across every source, and returns a sourced map. The query language is optional.
How it works
You ask; Jax queries across the world model, telemetry, and history, runs the spatial analysis, and renders the answer as a layer.
1You: "Which substations have more than 200 customers within a 10-minute drive,
2 and how has that changed over the last year?"
3 1. Query Jax writes SQL across the world model + live telemetry + history
4 2. Analyze isochrone (10-min drive) → spatial join → aggregate by substation
5 3. Compare difference against the same window one year ago
6 4. Render a styled layer + summary statistics, in your projectWhen a question needs optimization rather than analysis — routing, allocation, coverage — the GeoEngine hands it to the MathEngine and folds the result back into the same map.
Spatial operations
Jax composes the right operations for the question asked — you never name them.
| Category | Operations |
|---|---|
| Geometry | Buffers, spatial joins, nearest, distance matrices, clipping |
| Access | Isochrones and drive-time, reachability along the road network |
| Patterns | Density, clustering, hotspot detection across space and time |
| Aggregation | Roll-ups by geography, attribute, and time window |
| Optimization | Routing, allocation, and coverage handed off to the MathEngine |
Example
A network planner needs to find underserved areas before the next build cycle.
- Drop in a CSV of current service points and ask: "Where do we have more than 500 households outside a 15-minute drive of any existing node?"
- Jax builds 15-minute drive-time isochrones from every node, spatially joins household footprints from the DataStore, and aggregates the uncovered ones.
- The answer renders as a styled coverage-gap layer with a count and a hotspot overlay, sourced and reproducible.
- The planner asks a follow-up — "rank those gaps by household density" — and the layer re-colors in place.
- The strongest gap is forked into a branch and handed to engineering to design against.
Swap the layers and the same ask-query-analyze-render loop serves demand modeling, risk overlay, or terrain assessment. The spatial methods are generic; the meaning comes from what you analyze.
Where to go next
- Nexma GeoEngine — the spatial engine that runs every query.
- Nexma MathEngine — the optimization spatial analysis hands routing and coverage to.
- DataBase — the time-aware backbone behind every "how has it changed" question.
- Autonomous Engineering — turn a coverage gap into a design.
- Real-Time Digital Twin — run spatial analysis against the live world model.