You can analyze your department's NERIS data with AI by connecting your incident records to an AI assistant you already use, such as Claude, ChatGPT, or Copilot, and asking questions in plain language. With FlorianAI MCP, a chief can ask "Are we giving more mutual aid than we receive?" or "How often did we fight fire on tank water this year?" and get an answer computed from the department's own NERIS records, with a note on how complete the underlying data is. No export, no pivot table, no waiting on a report request. The connection is read-only, so your NERIS records never change.
What Can Fire Chiefs Learn From Their NERIS Data?
NERIS, the National Emergency Response Information System led by the U.S. Fire Administration, replaced NFIRS as the national incident reporting standard. Every incident your crews document feeds a structured record: incident types, unit responses, timestamps, aid given and received, fire detail, casualties, and more.
That record is far richer than most departments ever use. Inside it are answers to questions chiefs get asked every budget season:
- How much of our workload is mutual aid we provide to neighboring departments?
- Are we meeting the response objectives we have committed to?
- What does our water supply picture actually look like on fire calls?
- How many firefighters were injured this year, and on what kinds of calls?
- Does call volume actually drive our overtime, or is something else going on?
Each of those questions has an answer sitting in your NERIS data today. The hard part has never been collecting the data. It has been getting the answer back out.
Why Getting Answers Out of NERIS Still Means a Spreadsheet
For most departments, answering a NERIS question looks the same as it did under NFIRS. Someone exports incidents for a date range. Someone else cleans up the file, filters by incident type, and builds a pivot table. If the question touches a second system, like staffing or payroll, that data gets exported too and lined up by date by hand.
By the time the answer reaches the chief, it is days old and the question has often changed. Worse, the person building the spreadsheet is usually a battalion chief or company officer who has other work to do. Every ad hoc request pulls them off it.
There is also a quieter problem: data quality. NERIS fields are not always filled in. Aid records are sparse in many departments. Staffing on every arriving unit is often missing. A spreadsheet built in a hurry rarely tells you how much of the data was actually there, so a number that looks solid may rest on a small fraction of your incidents.
We covered the reporting side of this, getting incidents into NERIS cleanly, in our post on NERIS reporting automation with AI. This post is about the other direction: getting answers out.
How FlorianAI MCP Connects NERIS to Claude, ChatGPT, and Copilot
FlorianAI, an AI operations assistant built for fire departments, connects to your department's systems, NERIS included. FlorianAI MCP extends that connection to the AI assistant your department already uses. MCP is the open standard that lets an assistant like Claude, ChatGPT, or Copilot reach data held in outside systems.
Once connected, you ask your question the way you would ask a deputy chief. Behind the scenes, FlorianAI pulls the matching NERIS incidents for the period you asked about and computes the answer in code, not by having the AI guess at the math. The assistant then explains the result in plain language.
Three things matter for how a chief can rely on the answer:
- It is read-only. FlorianAI reads your NERIS data. It never creates, edits, or deletes an incident record.
- The date range is explicit. Ask about "last quarter" or "this fiscal year" and the answer states exactly which dates were measured, so there is no ambiguity about what was counted.
- Coverage comes with the answer. Each result reports how many incidents were in range and how many actually recorded the field in question. If only a fraction of your fire incidents recorded a water supply, you see that before you put the number in front of council.
NERIS Questions You Can Ask Today
These are the kinds of questions departments can ask now, each answered from NERIS records by a purpose-built analysis rather than a general-purpose chatbot.
Are we giving more mutual aid than we receive?
Ask about mutual and automatic aid over any date range and FlorianAI reports aid given versus aid received, your department's net posture as a provider or recipient, a breakdown by aid type, and the partner departments you exchange the most aid with. It also counts aid your department received from non-fire agencies such as EMS, law enforcement, and public works.
This is the question behind a lot of regional cost-sharing conversations. If your department is quietly covering a neighbor's calls, the data shows it. Because aid records are sparsely populated in NERIS, the answer tells you how many incidents carried an aid record at all, so you know how much weight the number can bear.
Are we meeting NFPA 1720 response objectives?
For volunteer and combination departments, FlorianAI assesses incidents against the NFPA 1720 demand-zone objectives. Each incident is placed in an urban, suburban, or rural zone by census population density. For each zone, you see first-unit response time from dispatch to on scene and the personnel assembled at the scene, measured against that zone's objective and required percentage.
If your department has adopted its own objectives through a standards-of-cover study, those can be used in place of the defaults. Zones with too few incidents to judge do not get a verdict, and the answer calls out incidents that could not be classified or were missing a response time or staffing count. Career departments working against NFPA 1710 can ask the equivalent response time compliance question. For a deeper look at response time analysis specifically, see how fire departments use AI for response time analysis.
What does our water supply look like on fire calls?
NERIS captures fire detail that NFIRS-era reports rarely made easy to analyze. Ask about water supply and FlorianAI ranks how often your fire incidents ran on tank water, hydrants above or below 500 gpm, drafting, tender shuttles, or no water supply at all. The same analysis covers suppression appliances used, from hand lines to master streams, and whether an investigation was needed.
For a department weighing a tender purchase or a hydrant improvement request, that breakdown is the start of the case.
How many firefighters and civilians were injured this year?
Casualty questions need to be exact. FlorianAI separates firefighter and civilian casualties, fatal and non-fatal injuries, and uninjured rescues, and shows which incident types produced them. Dead-on-arrival medical patients are reported separately rather than folded into fatality counts, which keeps the numbers honest when they reach a report or a board meeting.
Does call volume actually drive our overtime?
This is where connecting more than one system pays off. With both NERIS and your scheduling data connected, FlorianAI lines up incidents per day against overtime hours per day and measures how closely they move together. It can also shift the comparison by a day, because incident data and 24-hour tour data often date the same shift differently.
The answer is stated as an association, not proof of cause. If the relationship is weak, that is a finding too. It tells you overtime is being driven by something other than call volume, such as vacancies, leave patterns, or minimum staffing rules, and points the next question in the right direction.
What Makes AI Answers on NERIS Data Trustworthy?
A general-purpose chatbot handed a spreadsheet will produce an answer whether or not the math is right. That is not good enough for a number a chief will defend in front of elected officials.
The difference with FlorianAI is that the counting is done in code against defined rules. Every result states its counting rule, its date range, and its coverage. When the data is thin, the answer says so. When a standard only applies to certain incident types, the answer says that too. NFPA 1720's staffing objective, for example, is defined for the initial full-alarm assignment to a structure fire, and FlorianAI flags that distinction rather than applying it to every call.
The AI assistant's job is to understand your question and explain the result. The analysis itself follows fixed rules, so the same question over the same records returns the same answer every time it is asked.
Do You Need an IT Project to Get Started?
No. FlorianAI MCP works with the AI assistant your department already has, and FlorianAI handles the connection to your NERIS data. There is no new dashboard to learn and no data warehouse to build. Line personnel do not change anything about how they document incidents.
If your department already runs FlorianAI, reach out and we will get MCP configured. If you are evaluating FlorianAI and want to see these questions answered against real NERIS data, schedule a demo.
Frequently Asked Questions
Q: Can AI change or delete our NERIS incident records?
A: No. FlorianAI MCP is read-only. It reads your NERIS data to answer questions and never writes to, edits, or deletes an incident record.
Q: Which AI assistants work with FlorianAI MCP?
A: FlorianAI MCP connects to Claude, ChatGPT, and Copilot, so your department can ask NERIS questions from the AI assistant it already uses.
Q: How do I know if an answer is based on complete data?
A: Every answer reports how many incidents were in the date range and how many recorded the field in question. If coverage is thin, you see it alongside the result.
Q: Do we need to export our NERIS data first?
A: No. FlorianAI pulls the matching incidents for the date range you ask about. There is no export, spreadsheet, or manual cleanup step.