Skip to main content

Practical AI applications for Māori GIS

The most useful AI jobs in GIS are often ordinary, repetitive jobs across a lot of text. The model can find candidates, restructure material and draft working outputs. The source remains the evidence.

Search many reports​

Use the Search your documents workflow to ask questions across a collection, then follow results back to the original pages. This is useful for repeated references to an awa, land block, road, pā, kāinga or person-place relationship.

Find candidate ingoa wāhi​

Use Find ingoa wāhi to extract wording, source page and surrounding context. Then use Check the places before assigning coordinates.

Extract tables​

AI can help turn a table trapped in a PDF into CSV-shaped text. Check row counts, headers, dates and totals against the source before treating the result as data.

A useful prompt is:

Extract this table exactly as rows and columns.
Preserve original wording and numbers.
Add a final column called source_page.
Do not infer missing values. Leave them blank.

Compare versions of plans​

For two versions of a plan, policy or report, ask the model to identify changed sections and preserve the heading or page reference for each change. GIS can then focus on changes that affect locations, zones, projects or named places.

Draft metadata​

Give the model a dataset description, source page and field list, then ask for a first-pass metadata note covering:

  • source
  • date
  • purpose
  • fields
  • transformations
  • limitations

Review it against the actual source. See Provenance and source checking.

Clean structured text​

AI can help standardise obvious formatting such as date formats, repeated whitespace or category spelling. Keep the original field and create a cleaned field where the change matters to provenance.

Draft QGIS expressions​

A model can help write or explain a QGIS expression. Test it on a small subset and inspect the result before applying it to thousands of records.

Describe field photographs​

For non-sensitive project photographs, image-capable models can create candidate descriptions or tags. The field record should still retain date, location, observer and project context. A generic image description is not a substitute for what the observer knew about the place.

A complete pattern​

Conceptual flow

See the Archives to map worked example for the full sequence.

Last reviewed: 26 August 2026

Keep the applications tied to the kaupapa​

For whenua, ingoa wāhi and taiao work, the useful AI applications reduce repetitive reading or data preparation while keeping reports, maps and records as the evidence. Archives to map and Find ingoa wāhi show how that source-linked workflow continues into GIS.

Last updated on