I’ve been around the emergency management profession for a long time. To put it in perspective, when I started at the Ohio Emergency Management Agency (EMA) in 1994, only two people had cell phones, the director and the chief of staff, and they were the size of a brick…literally. The apex of information dissemination was a blast fax. For those who don’t know what that is, and I suspect that is many, you feed a press release or an announcement into a fax machine, and it sequentially dials preloaded fax machine numbers and transmits the document. It might have taken two hours or longer to dial every number; however, for me, it was a game-changer. I went from manually dialing and faxing press releases to sending them to 50 media outlets with one button. Although this “cutting-edge” technology feels primitive today, the challenge it solved is remarkably familiar. How do we help emergency managers process more information, coordinate faster, and make sound decisions when time and capacity are limited?
Whether you work in a one person (or half a person shop) or in a state emergency management agency with more than 100 staff members, you face the same demands to process information, coordinate partners, and make sound decisions, rapidly and often without the needed resources. Increasingly, emergency management agencies at all levels are realizing that artificial intelligence (AI) may help ease that burden. But successful adoption requires a thoughtful approach that considers not just the benefits but also the risks.
That is the central message of the AI for Disasters + Emergencies Initiative (AIDE), led by the Markle Foundation in collaboration with Aspen Digital and RAND. I also heard the same priorities while moderating the “Emergency Management Today” leadership panel at September’s AIDE Summit in Austin.
After leading Ohio EMA and serving as FEMA Region 8 Administrator, I look at technology through a practical lens. Will it help a team get the job done on an ordinary Tuesday, and will it still help when that same team is managing an emergency on Friday? The AIDE Initiative and its Practitioner Playbook give emergency managers a practical framework for answering that question.
What emergency managers need and what stands in their way
AIDE started with two questions: What do emergency managers need, and what does the market offer?
Aspen Digital gathered practitioner perspectives and identified 40 potential uses of AI, with situational awareness drawing the strongest interest, followed by planning, public information and warning, and operational coordination. These are information-heavy functions where help drafting, summarizing, translating, and organizing information could give staff valuable time back.
RAND independently assessed AI-enabled products and identified 1,179 available products. Many either serve broad commercial needs or focus on one slice of emergency management, requiring additional configuration or integration to be useful. Separate contracts, data requirements, integrating into current platforms, and training can overwhelm an agency, especially one with limited resources.
That gap matters. Emergency managers do not need more technology choices. They need tools that solve real operational problems and can work within their existing systems, staffing, data, and governance.
Meanwhile, some staff are already using personal AI accounts and limiting inputs to public information because organizational guidance is unclear. Application is moving faster than governance.
Where AI tools still fall short
RAND found that products cluster in response, appear moderately in recovery, and thin out in preparedness. Tools supporting coordination across organizations remain limited, as do those focused on needs such as sheltering, case management, and long-term recovery.
Practitioners pointed to grant administration, exercise development, after-action reviews, and knowledge retention as promising applications. However, those uses do not carry equal risk. For instance, AIDE’s Practitioner Playbook places developing exercises and after-action reviews in a lower-risk category than supporting grant applications and management. A tool can be easy to use and still require strong guardrails.
Caution is a sign of professionalism
Emergency managers have raised legitimate concerns about accuracy, privacy, security, and overreliance. That is not resistance to innovation. It is responsible leadership from professionals entrusted with life-safety decisions, public resources, and sensitive information.
Aspen Digital’s findings reinforce that successful AI adoption requires more than promising technology. It depends on human-centered and transparent practices, strong leadership, staff capacity, effective procurement, and the right IT and data infrastructure.
Putting the Playbook to work
The Practitioner Playbook addresses both organizational readiness and individual use cases, giving agencies a practical framework for building the governance, knowledge, and processes needed to use AI responsibly.
Its practical guidance comes down to a few fundamentals, which we follow at IEM:
- Start with the problem, not the product. Define the need, establish a baseline, and determine what success looks like before introducing AI.
- Keep people in charge. AI should support human judgment, with accountable officials retaining responsibility for decisions.
- Plan for real-world conditions. Test tools against imperfect data, connectivity issues, and other disaster conditions, and maintain fallback processes.
- Use data fit and watch for unintended consequences. Understand what data a tool relies on and test for gaps or biases that could affect outcomes.
For recovery programs, that also means establishing how AI-assisted content will be identified, reviewed, approved, and retained, particularly when records may support reimbursement claims or be subject to later review.
Five questions to ask before you buy
The Playbook’s vendor checklist offers a practical starting point:
- What specific problem will this tool solve, and how do we measure it today?
- Who owns the decision, the data, and the results?
- What data does the tool need, where is it stored, and who else can see it?
- What happens when the connection drops or the tool fails?
- What will it cost over three to five years, including integration, support, and exit?
The Playbook is part of a broader effort to help agencies learn from one another, build capacity, and adopt AI without starting from scratch.
Readiness is built before the warning goes out
Technology has changed dramatically since my blast-fax days. The fundamentals of emergency management have not. Experienced judgment, trusted relationships, or local knowledge still matter most. Used responsibly, AI can reduce routine burdens and give teams more time to put that expertise where communities need it most. That readiness is built before the warning goes out, with clear rules, practical testing, and people prepared to use the tools responsibly.
At IEM, we see the same challenge across emergency management, disaster recovery, and grants management. The real opportunity is not simply doing paperwork faster. It is giving emergency managers more capacity for the work that requires judgment, relationships, and experience while maintaining the oversight, documentation, and transparency public programs demand.
Is your organization ready to put AI to work responsibly? Connect with the IEM team to assess your readiness, identify practical use cases, and build an approach grounded in your operational needs. CONTACT US
To read the full AIDE reports and Playbook, visit aideinitiative.org.
Acknowledgment of AI Use
I acknowledge the use of Claude to help generate initial ideas and structure for this blog post.
Featured Case Study
