Agentic GIS is GIS driven by an AI agent. Instead of a person choosing tools, layers and steps, an agent built on a large language model reads a plain request, plans a spatial analysis, runs the steps and returns a map or a table. Asked which stands within 5 km of the mill have mature pine and no harvest in the last ten years?, it would find the layers, filter, buffer, intersect and report.
How well it works today
It is early. The most useful evidence comes from benchmarks with exact answers. GISAgentBench, published in August 2026, built 349 multi-step tasks from questions posted by GIS practitioners, each with a reference output. Across six models, "the best agent completes only 32.7% of tasks under strict tolerance-aware scoring, although most models produce outputs that are close to the ground truth".
An earlier benchmark, GeoAnalystBench from 2025, measured something looser: whether the workflow an agent wrote was valid. There, one proprietary model reached "high validity 95%", while a small open model managed 48.5%. Tasks "requiring deeper spatial reasoning, such as spatial relationship detection or optimal site selection, remain the most challenging across all models". A valid workflow can still give the wrong answer, which is why the two figures differ so much.
The 2026 tasks came from a question-and-answer site, where people post what they couldn't solve. Routine requests may go better.
Why near misses matter
GISAgentBench found most outputs "close to the ground truth" without being right. In spatial analysis, a near miss looks like an answer. A buffer in the wrong units or a join on the wrong field still produces a plausible map.
What it isn't
Agentic GIS doesn't change the data underneath. The layers, identifiers and data quality still decide whether the answer is right.
What to check
Test any agentic GIS tool on questions whose answers you already know, and check the steps it took as well as the output. Keep a person responsible for any map that feeds a contract or a compliance claim.
Quarri for forest management is built around how a forest operation runs, from the cruise to the settled account.
For more depth: What is agentic GIS?
Sources
- Pothuri, Jiang, Xu and Yang, "GISAgentBench: A Practitioner-Sourced Benchmark for Evaluating LLM Agents on GIS Tasks", arXiv 2608.01645, 3 August 2026: arxiv.org
- Zhang, Gao, Wei and others, "GeoAnalystBench: A GeoAI benchmark for assessing large language models for spatial analysis workflow and code generation", arXiv 2509.05881, 7 September 2025: arxiv.org
Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.