What Is Spatial Intelligence? From Fragmented Data to Explainable AI

"Spatial intelligence" gets used loosely: sometimes it means a map with more layers, sometimes it means location analytics, sometimes it's just a rebrand of GIS. Underneath the buzzwords, though, there's a real and useful distinction from ordinary location data or dashboards.

Spatial intelligence is what you get when place-based data (documents, drawings, sensor readings, spreadsheets, regulatory constraints) is linked into one model that understands how those things relate to each other, and can be interrogated in plain language with an explainable, traceable answer.

01 Beyond Maps and Dashboards

Location intelligence and GIS tell you where something is. A dashboard can plot assets, sites, or sensors on a map and let you filter and zoom. That's useful, but it stops at the geometry. It doesn't know that this asset belongs to that project, that this reading contradicts a historical record three years old, or that this regulation applies here but not two fields over.

A spatial intelligence platform adds the relationships. It's the difference between a map with dots on it and a model that understands how those dots connect: which entities belong together, which records supersede others, and which constraints interact, all grounded in real spatial and organisational context.

02 Why Generic AI Struggles With Place-Based Data

Off-the-shelf chatbots are confident by default. Ask a general-purpose AI assistant a specific question about a site, an asset, or a regulatory constraint, and it will usually answer, whether or not it actually has grounded information to answer from. Without a structured source of truth behind it, that confidence is a liability, not a feature.

For decisions with physical, financial, or regulatory consequences, "confidently wrong" is worse than "clearly uncertain." That's why explainable AI matters specifically in this space: an answer is only useful if you can see where it came from.

03 The Building Blocks: Knowledge Graphs and Explainable AI

In practice, a spatial intelligence platform is built in four stages. Data is ingested and normalised regardless of source format (documents, CAD, spreadsheets, GIS files), so nothing has to be manually wrangled before it's usable. That data is then linked into a knowledge graph: a structured model connecting entities across space and time, rather than a pile of disconnected records.

Analysts and AI agents then interrogate that graph directly, with every answer grounded in it rather than guessed, which is what eliminates hallucination. Support for open standards like MCP means the same graph stays usable as the agents and tools built on top of it change. And because the graph holds an organisation's real data and expertise, keeping it sovereign and secure isn't optional; it's the whole point.

04 What This Looks Like in Practice

The same underlying model applies to very different problems. Compass: Terra turns borehole logs and ground investigation reports into a queryable site model. Compass: Locus applies the same approach to place-based delivery strategy, from climate adaptation and transport to energy-network connection. Compass: Res builds living portfolio intelligence across an entire property or asset estate. Where none of those fit exactly, Compass: Custom configures the same engine to a specific organisational challenge.

Different domains, same underlying shift: fragmented, siloed data becomes one coherent, explainable model that anyone in the organisation can query directly.

SimAnalytica builds spatial intelligence infrastructure for organisations working in the physical world.