FireMic for software vendorsFire department, not a vendor? The firefighter site is here.
For RMS, CAD and fire-service software
A firefighter taps a mic in your incident form and talks. Seconds later the form is filled, ready for review. You add the button. We build everything behind it, in your stack.
The demo takes under a minute, no sign-up. Sandbox API keys the same day.
Filled from the narration, marked for the firefighter to check. Your form, your save button, your filing.
A spoken call becomes a filled report, and the report gets filed. No sign-up. At the end, try it with your own voice.
A sample incident. The filing at the end is simulated, and nothing you change is saved.
February 1, 2026
NERIS replaced NFIRS. Every department that had a reporting routine needs a new one.
89%
Of departments reporting a type are volunteer or combination. The people with the least patience for a form.
30,740
Fire departments in the federal directory, each addressable by its NERIS id on day one.
In the big records systems, voice is a paid add-on. Where it exists it is priced separately, and a department that has not bought it has nothing to talk to. If that line on your price list is empty today, this fills it.
Counts of fire departments in the country, from the NERIS public directory as of July 28, 2026, not customer counts.
In your incident form. A menu item or a shortcut works too.
Your server. Or, with no server tier, a key pinned to that install.
One message from the hosted mic, or one HTTP call if you record yourself.
The whole pipeline our own app runs, as one call.
Written in your language against your field list, with tests, sent as a patch.
Desktop, iOS, Android or web, adapted to how your app opens a window.
What the narration missed comes back named, with the question to ask.
A better model ships on our side. Your build does not change.
Shortest
One page of ours in a webview or an iframe. Recording, upload and the waiting screen happen inside it. It hands you the draft.
const widget = FireMic.mount(el, {
sessionToken,
onDraft: (n) => fillIncidentForm(n.neris, n.transcript, n.gaps),
});Host code for every platform is in the embedding guide.
Most control
Record however you already do and post it once. Back comes the transcript, the draft in the shape NERIS accepts, and what is still missing.
curl https://app.fire-mic.com/v1/narrations \
-H "Authorization: Bearer $FIREMIC_KEY" \
-d '{ "department": "FD34007744",
"audio": { "mime": "audio/mp4", "data": "<base64>" } }'Every endpoint and an OpenAPI spec are in the API reference.
Round trip for a one-minute narration is 4 to 8 seconds, measured against production in September 2026.
The plan for a new partner. It needs an engineer for a few hours and one department willing to pilot.
Thirty minutes. Send us a field list or a sample export.
Mapping and host code, in your language, as a patch on a branch you name.
Your engineer merges. We narrate real calls together and compare field by field.
Live key, outcome reporting on, and we watch the first narrations with you.
Or ask for a sandbox key and start on your own.
This API has no submit route. You stay the submitter of record, with your validation and your audit trail in front of every filing.
No FireMic accounts, no login, no interface of ours in the loop. The department is identified by its NERIS id.
Never shared with another partner. Never used to train anything outside FireMic. Retention is a setting you choose.
No. There is no submit route on this API and it cannot reach ours. You stay the submitter, with your certification and your validation in front of every filing.
A button that opens our mic with a session token, and a handler for the draft it sends back. Or one HTTPS call if you record natively. The mapping from our draft to your fields is code we write for you.
Yes. The mic runs in WebView2, and a key pinned to one department lets that install mint its own sessions. A key from one station cannot act for another.
Good enough that our own app files with it. We benchmark against a fixed set of narrations before any model change, and for a partner we report accuracy per department from the outcome data you send back, denominator attached.
Yes. A sandbox key is issued the same day you ask and runs against the same production pipeline. Live keys follow the agreement.
Partner terms are per agreement. The sandbox costs nothing and needs no agreement, so you can find out whether it works before anyone talks price.
A field list or a sample export is enough to start.