SYSTEM LOG // VERIFIED
Don’t Believe Me?
OptiCall isn't real. I made it up.
Every capability it has, though, is already running on a real city's 911 system in some fashion or another. And who knows, possibly where you live.
The systems and algorithms are listening to your call. Scoring your value. And quietly routing you to the cheaper outcome instead of the fastest one. None of this is speculative fiction -it's being built into the current U.S. emergency response infrastructure by real companies, funded by real money, running in real cities right now. Here are six of them, with sources, so you can go check my work yourself.
SEATTLE, WASHINGTON
AI was listening on 911 calls - and no one told the callers.
Starting in December 2023, Seattle Fire Department let a Danish AI company named Corti monitor 911 medical calls in real time, flagging which callers should get diverted to a nurse call center instead of an ambulance. It ran for more than two years with zero public disclosure -no notice to callers, no published data on accuracy or bias. Diversions reportedly jumped 32-50%. The nurse line already had a caller die after a ten-hour wait.
NATIONWIDE
Your ambulance might answer to a private equity firm
Global Medical Response -the largest ambulance operator in the country -has been owned by the buyout firm KKR since a $2.4 billion deal in 2017. As of early 2026, KKR is preparing it for a U.S. stock listing. The crew that shows up when you call 911 may work for a company whose real customer is a shareholder.
NEW YORK CITY
The algorithm that cried gunfire.
NYPD's ShotSpotter system uses AI-tuned microphones to detect gunshots and auto-dispatch officers. A city comptroller audit found 87% of its alerts turned up no confirmed shooting -only 13% did. Officers burned thousands of hours chasing false positives, for a real speed advantage over a 911 call of about ninety seconds, not the five minutes it's marketed on.
PORTLAND, OREGON
A software update put people on hold while they died.
When Multnomah County switched its 911 centers to new dispatch software in May 2021, five-plus-minute hold times shot up nearly 500% in the first month and over 800% within nine. By March 2022, only 40% of calls were answered inside the 20-second national standard, down from 87% two years earlier. One caller described being bounced on hold for fifteen minutes while her husband lay unconscious.
MULTIPLE STATES
The dispatch software nobody voted for.
That same family of dispatch software has been blamed in at least three states for serious errors -including crashing mid-call and contributing to a botched criminal case. Minneapolis dropped it in 2019, saying it wanted dispatchers "to talk and think for themselves" again.
EVERYWHERE
Your zip code still decides how fast help arrives.
Peer-reviewed research cited by the CDC shows ambulances respond measurably slower to poorer neighborhoods than nearby wealthier ones -not purely a matter of distance, but of how call volume, staffing, and routing get modeled. The math doesn't have to intend discrimination. It just has to keep landing on the same zip codes.
None of this required a conspiracy. It just required a spreadsheet, a cost center, and nobody asking the second question. That's the DNA of my book, First, Do No Harm.
Sources linked above are independent reporting and public audits, current as of publication. Laws, contracts, and vendors change -verify current details before truly believing or citing.