AI Needs GIS
By John Brosowsky, SVP of Innovation, GeoComm
Artificial intelligence is rapidly reshaping PSAP operations. From handling non-emergency calls to enabling real-time 9-1-1 transcription, translation, keyword detection, and call taker assistance, AI is becoming an integral part of modern emergency communications.
But AI is not inherently intelligent; it identifies patterns and relationships within the information it has been trained on and the data it can access. Its ability to answer questions, make predictions, or generate recommendations is constrained by that information. One important source of information for AI is GIS data.
State and local governments maintain extensive GIS data covering street addresses, transportation networks, NG9-1-1, public safety, emergency response, buildings, property records, critical infrastructure, utilities, public works, asset management, schools, parks, public facilities, environmental data, natural hazards, imagery, elevation data, and hundreds of other authoritative location-based datasets.
More than just “layers on a map” to look at, this GIS data is knowledge. It's knowledge about space and place. Knowledge that is often known only at the local level and is often private and secure. Knowledge that, if known during an emergency response, could help save lives and protect property. Unfortunately, this knowledge frequently goes unused because 9-1-1 call takers, dispatchers, and emergency responders don’t have time to search through hundreds of map layers during an emergency.
But connecting this knowledge to PSAP and responder AI systems solves this problem. GIS data gives AI location intelligence, enabling it to know where things are and to make important spatial connections in real-time during 9-1-1 calls.
Technical Signals
Esri released NG9-1-1 Data Manager version 2.5, with three additions relevant to NG9-1-1 GIS data preparation. A bulk road centerline tool updates address range fields across multiple segments in a single operation. A new Multiple Dwelling Unit editor handles unit-level addressing for multi-story buildings. Custom lifecycle workflows let administrators control which promotion stages are available to different user classes, tightening quality gates before records move toward submission. The tool remains aligned with NENA's NG9-1-1 GIS Data Model v2A, with version 3.0 targeting the updated data model later in 2026.
Geospatial AI teams need to match imagery resolution to the scale of the objects they want models to identify. Medium-resolution satellite imagery can support analysis of forests, rivers, agricultural fields, and broad land-cover classes, while vehicles, rooftops, utility assets, and people generally require much finer imagery. Higher resolution also increases acquisition costs, storage needs, dataset sizes, and processing time. Effective GeoAI workflows should first define the target feature, confirm that it is represented by sufficient pixels, assess its spectral and temporal characteristics, and then select and validate an appropriate model.
More than 1,000 U.S. public safety agencies now hold FAA Part 91 BVLOS approval to operate Drone as First Responder DFR) programs, following a streamlined waiver process introduced under the FAA Reauthorization Act. Modern DFR platforms combine thermal imaging, automated license plate recognition, and Real Time Crime Center integration to improve emergency response. The proposed FAA Part 108 rule, now under final White House review, would establish a permanent national framework for BVLOS operations. For GIS and public safety agencies, the shift is expected to accelerate drone adoption while increasing the need for robust airspace coordination, dispatch routing, and aerial data governance.
Zetron demonstrated an AI workflow at Critical Communications World 2026 in London that generates a structured risk assessment from an active incident and pushes it from the emergency call center to officers in the field before they arrive on scene. The capability was developed through a trial with a UK public safety agency and extends AI output beyond the call-taking screen into the field response workflow. As AI tools extend from single-screen call support to cross-workflow intelligence, the address point, building record, and incident boundary become the shared reference that connects what the call center sees with what the field team receives. Gaps in that location foundation degrade the AI output at both ends of the handoff.
GIS Policy & Standards Watch
The National Emergency Number Association has published NENA-STA-011.2-2026, an ANSI-approved standard for 9-1-1 professional education developed over five years. The standard provides institutions with a common framework for creating or improving education programs for 9-1-1 professionals and maintaining consistency across curricula. NENA credited its 9-1-1 Professional Education Working Group with developing the final standard.
Massachusetts launched its NG9-1-1 Address Finder statewide, giving municipalities, GIS professionals, emergency communications personnel, and the public a direct way to validate addresses against the authoritative dataset used to route 9-1-1 calls. Users search for an address, check whether it is correctly represented in the state's Master Address Database, and submit a correction report when something appears to be missing or incorrect. For GIS coordinators working toward NG911 address readiness, Massachusetts demonstrates what a statewide feedback loop looks like when it connects directly to the data used to route emergency calls.
NENA published its Standard for 9-1-1 Notifications and Coordination with the U.S. Coast Guard, establishing standardized procedures for PSAP coordination during maritime emergencies. The standard formalizes how 9-1-1 centers handle water rescues and other coastal incidents, replacing inconsistent local practices with a common framework. It highlights the importance of accurate shoreline boundaries, maritime coordinates, and jurisdictional mapping to support seamless call transfers and coordinated emergency response.
Insight of the Week
The Fairfax County Police Department's Drone as First Responder program received the National Association of Counties 2026 Best in Category award for Criminal Justice and Public Safety. Operated through the department's Real Time Crime Center, drones reach scenes ahead of ground units in approximately 70% of deployments, on average, within two minutes. The program recently added AI to evaluate 9-1-1 calls in real time, triggering dispatches faster. The program is also the first in the National Capital Region to receive FAA authorization for beyond-visual-line-of-sight operations in restricted airspace.
Resources & Events
Report Spotlight: Counter-UAS Interim Final Rule for State, Local, Tribal, and Territorial Agencies (DHS and DOJ)
The Department of Homeland Security (DHS) and the Department of Justice (DOJ) published an Interim Final Rule implementing the SAFER SKIES Act, giving state, local, tribal, and territorial law enforcement and correctional agencies independent authority to conduct counter-unmanned aircraft system (C-UAS) operations protecting large public gatherings, critical infrastructure, and correctional facilities. The rule took effect July 1, 2026, and establishes a two-tier certification structure. A free one-hour online course through the Federal Bureau of Investigation's National Counter-UAS Training Center certifies agencies in detection and warning operations, while mitigation actions, such as jamming a drone or seizing control of one, require resident training at the center. Agencies must use technologies from federally maintained authorized lists, obtain Federal Communications Commission approval before operating any system that emits radio waves, notify air traffic control in real time during mitigation, and report every mitigation action to DHS and DOJ within 48 hours. The Departments cite more than 600 drone incursions into restricted airspace during the opening weeks of the 2026 FIFA World Cup and expect about 1,500 agencies to certify at the detection tier within two years. Comments on the rule are due September 4, 2026.
Share or Forward
If you liked this edition of The Response Brief, please forward it to someone in your network who would also find value from this newsletter.


