Local governments manage vast amounts of information, from parcel boundaries, taxation records, zoning regulations, flood zones, easements, to liens and more. Yet this information rarely lives in a single system, and therein lies the problem. For GIS teams, the challenge is not just mapping this information but integrating it with ArcGIS and making it analysis-ready without disrupting existing systems of record.
At this year’s Developer and Technology Summit plenary, Shreyas Shinde presents three strategies for integrating scattered and disparate data with ArcGIS and transforming it into actionable insights.
Strategy 1: Register and Publish Data
For many local governments, data is already managed within an enterprise geodatabase. This is also the case for the Harris County parcel data Shreyas uses to demonstrate the first integration strategy.
To make the parcel data accessible throughout his organization, Shreyas registers the geodatabase with ArcGIS and publishes the data as map and feature services (map services with feature access enabled). The data is managed using the parcel fabric data model, which not only stores parcel geometries but also preserves the rules and relationships that govern how parcels are edited and maintained. After publishing, Shreyas adds the parcel and easement layers to a map, providing him and others in his organization with a spatial foundation for analysis and decision-making.
However, not all data reside in a geodatabase, which leads to Shreyas’ second integration strategy.
Strategy 2: Access External Databases with Custom Data Plugins
In many counties, taxation records are maintained by the county treasurer in their own databases and typically exist as non-spatial tabular data. Rather than copying this data into ArcGIS, Shreyas demonstrates how organizations can access it directly using Custom Data Plugins.
Custom Data Plugins are an extensibility pattern that allows developers to create feature services that reference data not natively stored in ArcGIS.
Shreyas shows his Node.js–based Custom Data Plugin that connects to a MongoDB database used here to represent a typical county treasury system. His plugin iterates through records in the database and dynamically converts them into GeoJSON features. The plugin, packaged in a custom data package file (.cdpk), is deployed to ArcGIS Server, where feature services can be created that reference the underlying data. The result is seamless access to taxation records within ArcGIS without duplicating or migrating the original data.
Custom Data Plugins provide a powerful way to integrate external data while preserving the integrity of the original data sources. But for situations where organizations need to periodically ingest and transform data from cloud storage, Shreyas introduces ArcGIS Data Pipelines and his third and final strategy.
Strategy 3: Use Data Pipelines for ETL Workflows
To integrate Harris County’s flood zone data stored in Azure Blob Storage with ArcGIS, Shreyas uses ArcGIS Data Pipelines to build a pipeline that retrieves the data and copies it into ArcGIS as a hosted feature service. The pipeline can be executed on demand or scheduled to run automatically, ensuring his organization always has access to the most current version of the flood zone data in the cloud storage.
With parcel data, taxation records, and flood zone information now accessible as feature layers and tables in ArcGIS, Shreyas and his team can begin mapping them to answer complex questions.
Traditional workflows often rely on multiple table joins and spatial queries. While effective, these approaches can become complex as relationships between datasets grow. To address this challenge, Shreyas demonstrates how integrated datasets can be transformed into a knowledge graph.
Shreyas uses ArcGIS API for Python to extract entities such as parcels, easements, tax records, and flood zones from feature services and store them in a knowledge graph service.
Once created, the graph can be explored using ArcGIS Knowledge Studio, where analysts can visualize the relationships between entities and run graph queries to better understand how different datasets connect, as Shreyas illustrates in his demo.
The final step in the demo introduces an emerging capability: using an AI agent to query the knowledge graph using natural language.
Within an ArcGIS Notebook, Shreyas configures a Python-based agent that has access to a large language model (LLM) and tools to discover and query the knowledge graph. This allows analysts in his organization to ask questions in plain language, such as:
- “Find all parcels that have an easement and that have a lien against them.”
- “Find all commercial parcels that are in the flood zone.”
The agent translates these requests into graph queries, retrieves the relevant entities, and returns parcel IDs that can be visualized on a map. This approach makes complex analytical workflows far more accessible to staff across his organization, not just GIS specialists.
It also opens the door to more advanced automation, which Shreyas demonstrates by creating a query that, based on edited parcel data, identifies all downstream entities and the departments in his organization affected by that change.
Local governments will continue to rely on multiple systems to manage their operational data. The key is not consolidating everything into a single platform, but enabling those systems to work together. By combining the strategies that Shreyas shared, organizations can bring their scattered data into ArcGIS for unified analytical workflows. Layering knowledge graphs and AI agents on top of this integrated data transforms them into connected knowledge, empowering analysts, planners, and decision-makers to ask better questions and uncover deeper insights. How will you implement these strategies in your organization?
Commenting is not enabled for this article.