What is Geospatial General Intelligence?
Geospatial general intelligence is a multi-agent AI system that ingests, processes, and reasons about any geospatial data — aerial, oblique, LiDAR, satellite, and street-level — to survey-grade accuracy, and verifies its own work with a provable check-point report. It runs the full production chain, from raw-data ingestion through to a published, quality-checked product. AG2I — Artificial Geospatial General Intelligence — is a working implementation.
Multi-agent geospatial intelligence
Multi-agent geospatial intelligence is geospatial production run as a coordinated team of agents rather than a monolithic pipeline. An orchestrator plans the work and adapts to failures; deterministic specialist agents do the compute — aerotriangulation, LiDAR, orthorectification, automated feature extraction; a verifier gates quality at every stage; and a shared blackboard is the single source of truth. The guiding principle is reasoning at the edges, deterministic compute in the core — a language model never sits inside a bundle adjustment. See the case study: A Mapping Company in Software.
How it differs from geospatial intelligence (GEOINT)
Geospatial intelligence (GEOINT) is the discipline of analyzing imagery and geospatial data. Geospatial general intelligence is the AI system that performs the entire production — ingestion, photogrammetry, surface modeling, orthorectification, feature extraction, quality assurance, and publishing — automatically and verifiably, as a team of agents. One is the field of analysis; the other is the system that does the work and proves the result.
What makes it "general"
Generality is in the breadth of data and the autonomy of the workflow: one system that conditions and processes any geospatial input across sensors and scales, chooses its own parameters per dataset, checks its output against independent ground control, and adapts when a result falls out of tolerance. The same Task Contract exposes every capability, so the orchestrator composes photogrammetry, LiDAR, ortho, and feature extraction into one pipeline without special-casing.
Evidence
AG2I's approach is documented in engineering case studies with measured accuracy against manufacturer references — independent aerotriangulation and orthorectification, zero-annotation HD-map feature extraction, and the multi-agent orchestration that runs production end to end and gates its own quality.
AG2I — Artificial Geospatial General Intelligence. ag2i.ai · info@ag2i.ai