Somewhere in Houston, a pipeline engineer is staring at a stack of 1970s-era paper maps. Each one shows a different segment of a gas transmission network—hand-drawn, coffee-stained, indispensable. Digitizing them used to mean weeks of manual tracing, one polyline at a time. Not anymore.
Esri's ArcGIS Pro 3.7, released in May 2026, introduces tools to automatically extract features from scanned maps, along with the ability to control layer visibility by map frame in layouts , according to the company's release notes. But the headline feature is something more fundamental: two new geoprocessing tools in the Conversion toolbox bring AI-assisted digitizing to ArcGIS Pro, dramatically reducing the amount of heads-up digitizing required , Geospatial Training Services reported. What once took hours now seeds itself in minutes.
The timing matters. Utilities and energy networks are under pressure to evolve fast, and geospatial intelligence is positioned at the heart of this transformation, enabling real-time visibility, predictive control, and strategic planning , according to the Geospatial World Forum 2026 agenda. Natural gas prices slipped to $3.25/MMBtu per MMBtu on Wednesday, down -2.40%, per market data—but infrastructure planning cycles stretch across decades, not trading sessions. The tools to map that infrastructure, however, are accelerating.
Can You Model a Pipeline Network Without Enterprise?
For years, the answer was no. ArcGIS Knowledge graphs previously required an ArcGIS Enterprise deployment to create and use, but in 3.7, you can now create and manage a file knowledge graph stored in a local folder—no Enterprise required , Geospatial Training Services noted. A file knowledge graph is a file in a folder on disk with a .knowledgegraph extension, using a SQLite database and supporting the full graph data model , Esri's documentation states.
This is not a minor tweak. Knowledge graphs model relationships—between wells and gathering lines, between substations and transmission corridors, between permits and parcels. A knowledge graph allows you to create a model that simulates a real-world system in a nonspatial manner, with information structured around entities and the relationships between them , according to Esri. Previously, building one required server infrastructure, IT approvals, and budget. Now it requires a folder.
This lowers the barrier significantly for smaller organizations, standalone GIS analysts, and consulting scenarios where an Enterprise deployment isn't practical , the training firm observed. For a midstream operator mapping right-of-way conflicts, or a mining consultant tracking drill core relationships, that barrier was the difference between "we should do this" and "we can't afford to do this."
What Does AI Actually Digitize?
Not everything. The new tools work on binary raster images—black lines on white backgrounds, the kind produced when you scan an old plat map or engineering drawing. Extract Scanned Lines and Extract Scanned Polygons automatically generate vector features from binary raster images of scanned maps , according to Geospatial Training Services.
The workflow is straightforward. An Optical Character Recognition (OCR) model in the Living Atlas identifies text areas within the scanned map using the Detect Object using Deep Learning tool, creating polygon features containing the extracted text , Esri's blog explained. Then edge detection kicks in. The Edge Detection model in the Living Atlas identifies the edges of features to be extracted using the Classify Pixels Using Deep Learning tool , the company added.
The output isn't perfect. It still requires cleanup—smoothing, simplification, trimming. But for agencies or organizations still working with historical map archives, what previously required hours of manual tracing can now be seeded automatically and refined as needed , the training service noted. In the energy sector, where legacy infrastructure often predates digital record-keeping, that's the difference between a six-week project and a six-day sprint.
IQGeo's AI-powered geospatial software mobilizes data from GIS and enterprise asset management systems into an application that works both online and offline , Adrian McNulty, the company's Vice President of Utility Solutions, told Renewable Energy World at DTECH 2026. Many utilities are already taking hundreds or thousands of images of their assets during inspections or outages, but they're failing to effectively leverage the data in those images , McNulty said.
