Local Crime Fighting Gets Anduril-ized: How Tech Startups Are Changing City Police Forces
Technology like surveillance drones and Palantir-style software is now being sold to public safety agencies of all sizes—prompting more questions about its ultimate use.
Recently, Nick Noone, co-founder and CEO of Peregrine, was introduced to Candace Parker, sports broadcaster and retired WNBA star, by mutual friends. They hit it off. The two share a passion for athletics. (Noone was a Stanford gymnast.) And both have an interest in startups. (Parker is an active angel investor.) After Noone described how his startup sells Palantir-esque data analytics software to local police and public safety departments, Parker insisted that he meet someone she knew: Shaquille O’Neal.
“Well, she’s on TV with Shaq,” Noone explained. “And she told me, ‘Shaq loves public safety.’” After she put them in touch, O’Neal and Noone spoke over FaceTime. A few weeks later, Shaq called Noone back. “It was, like, 6 p.m.—I’m out on a date at dinner—and I get a FaceTime from Shaq, and he’s like, ‘Yo, you got to meet the police chief of Baton Rouge: he’s the man,” said Noone, doing a fair Shaq impression.
He was recalling the story from his Baton Rouge hotel room one afternoon last month, having flown to Louisiana to meet the police chief earlier in the morning after O’Neal connected them. They didn’t strike a deal that day, but he left feeling optimistic. “It was a great preliminary conversation,” said Noone, having long honed a slow and steady sweet talk for winning over government bureaucrats, a skill he first developed while working at Palantir.
Noone fully expects the Baton Rouge chief will sign. “In the next, like, three to six months,” he said. He anticipates the chief will put in a call to his peers with the New Orleans Police Department, which already uses Peregrine; such existing customers prize Peregrine’s products for making it easy to search department records and data through simplistic, Google Search–style prompts, among other features. And besides, Noone, 37, can speak with the confidence of someone who has watched his company’s revenue climb more than sevenfold in two years to $75 million in 2025.
“There is just so much money out there for this,” he said.
That’s something of an understatement. America is experiencing a full-on boom in startups developing technology meant to combat crime, track crime, study crime, reduce crime, analyze crime—and many other tasks related to public safety. Venture capitalists are hugely eager, too, and have tossed those companies over $2 billion in the last two years, more than quintuple the amount from the prior two years. In Peregrine’s case, the startup received a roughly $400 million valuation last year—and then a $2.5 billion one in March from a fundraising led by Sequoia Capital.
At least two dozen such companies have joined the burgeoning field, finding willing customers in cities of all sizes across the country—ranging from major metros like Atlanta, Los Angeles, Las Vegas and San Francisco, to midsize cities such as Albuquerque, N.M., and Newark, N.J., to hundreds of small towns.
Drones are some of the most popular products, and news footage of drones in action regularly goes viral on social media as the country adjusts to an eerie new era of increasingly high-tech local police forces.
Many other startups sell software for digitizing parts of public safety that haven’t changed much in recent decades. RapidSOS, for instance, sells AI-powered software for handling 911 calls and other emergencies, while Kodex has a web platform, popular with public safety agencies and private businesses alike, that simplifies how companies track and respond to government subpoenas and other requests for user data and information. Dozens of firms—ranging from Fidelity and AT&T to OpenAI and Coinbase—use it.
Kodex co-founder Matt Donahue, 33, got the idea for the company after several years at the FBI, where he sometimes needed to fax such information requests. “I just got married, and when I was telling my wife’s relatives what I do for work—and I would explain what Kodex is—they’re like, ‘Wait, something like that doesn’t already exist?’” Donahue recalled.
Very few of these businesses sell directly to consumers, and the ones that have tried to do so have found pretty limited demand: Most people want the police and other parts of the government to maintain public safety—they don’t want to pay extra to do so themselves.
Rather, the vast majority of such startups want to do exactly what Palantir did with spies and federal law enforcement agencies and what Anduril did with the military: Win lucrative, recurring local government contracts, wedging themselves into markets Silicon Valley once mostly totally ignored, and take advantage of a culture shift within the technology industry, which no longer looks at purveying goods to the Man with disinterest and distaste.
Think of this period in time as I’ve come to: as the Anduril-ization of crime fighting.
“Palantir and SpaceX had to crawl, so that Anduril could walk, so that these new companies can run,” said David Ulevitch, a general partner at Andreessen Horowitz, which has funded many of these startups (and Anduril, too). Of course, Andreessen Horowitz isn’t alone: Many of the major venture capital outfits have piled in, including Founders Fund and Greycroft.
“There was broad social stigma against supporting the police,” Ulevitch said. “And people just didn’t think these were big markets, and now they’ve realized they’re massive markets.”
While Silicon Valley technologists may be more comfortable with public safety tech, the proliferation of such software and hardware has led to mounting worries that America is turning into a highly digitalized police state in local communities. The shift follows the longstanding adoption of similar technology among federal agencies.
Jay Stanley, a senior policy analyst with the ACLU, fears Silicon Valley is handing local police what amounts to an ecosystem of mass surveillance. “And mass surveillance is not a good thing,” he said. “Law enforcement should not proceed collecting information on everybody all the time: Just because we cando something doesn’t mean we shoulddo something.”
The new technology’s nationwide spread is ironic given how much of it comes from one very specific place: the Bay Area. Indeed, the initial groundswell of interest around it partly stems from the tech elite’s concerns about crime close to home. San Francisco, in particular, has had a very public battle with crime and drug use in recent years, leading to the recall of San Francisco District Attorney Chesa Boudin in 2022 and the election two years later of San Francisco Mayor Daniel Lurie, who has trumpeted his efforts to decrease crime—particularly in neighborhoods like Union Square and the Financial District.
When I spoke to venture capitalist Cyan Banister, who has funded several public safety startups, she described sensing a “vibe shift” among investors and founders toward increased policing in the Bay Area—and investment in public safety startups. “We don’t want to live with prison bars on our windows,” she said.
Sitting in a glass-walled conference room in San Mateo, Calif., Adam Bry was discussing how a technology once mostly associated with warfare—drones—is becoming an important part of police departments’ crime-fighting toolbox when a quadcopter drone whizzed past the window behind him. Very shortly, the drone—made by Skydio, the company Bry, 39, runs—vanished out of sight.
Skydio keeps several hundred of its flagship X10 drones across its campus, with many installed in rooftop charging pods. At any point in the day, a good number of them are cruising around the company’s property as its engineers test new features. When I visited Bry, I found they had a tendency to appear when I was least expecting them.
“Ohhh,” Bry said when I mentioned it. “You get used to it.”
Police in places like New York, San Francisco and smaller cities use the drones to conduct overhead surveillance. (Meanwhile, corporate customers such as utilities and construction firms buy them, too, to examine their facilities from above.) In New York, for instance, police have apprehended subway surfers—people riding atop moving metro cars—after spotting them with a drone. And in San Francisco, officers have relied on drones to follow car thefts as they happen—as well as nab a person who stole $15,000 in goods from the Union Square Burberry store. The X10 drones can fly autonomously, or someone can steer them manually via a device purposefully made to resemble a videogame controller.
These drones represent Skydio’s renewed chance at life. Like several other companies selling public safety tech, Skydio started out as a consumer business. But the company, co-founded by Bry and his MIT classmates Matt Donahoe and Abraham Bachrach in 2014, couldn’t outmuscle its Chinese competitors, especially Shenzhen-based DJI, the leading maker of consumer drones. (Police departments also use DJI’s drones.) So Skydio began to shift its focus to selling drones to public safety agencies and private companies in 2020, ending its consumer business entirely in 2023.
A new Skydio drone, the R10, has been designed to prioritize indoor flight, which allows police to scout an active crime scene without sending in any humans. The R10 will further turn up the company’s rivalry with Seattle-based Brinc, run by Blake Resnick, a 25-year-old former Thiel Fellow, who started the company eight years ago. Brinc’s specialty is its Lemur drone, also specialized to fly indoors. The aircraft can push open a door, break glass and transmit live video, audio and a lidar map of a building.
“So police know exactly what they’re responding to,” Resnick said.
All U.S. drone companies could get a real boost from a relatively little-noticed turn of events in Washington: Starting next week, DJI and another Chinese drone company, Autel Robotics, could face a ban on new sales in the U.S., stemming from a complicated bit of government machinations that involve both the Trump and Biden administrations.
“This will completely change my business,” Resnick said. “It’ll completely change every drone manufacturer’s business.”
Drone footage is just one of many forms of video, text and data that can get sucked into the software made by Nick Noone’s Peregrine.
He initially developed the plan for the startup while spending five years at Palantir. There, he worked within a small group that found itself frequently reassigned based on the company’s current most urgent problem—whether that involved internal budgets or frantic phone calls from a retail client that had discovered human trafficking within a part of its business. (Both were actual issues the unit faced, Noone insists.) After a while, Noone realized Palantir’s complicated and expensive software limited its pool of customers.
“It’s a whale hunt,” he said.
Noone quit Palantir in 2017, figuring he could make simpler but similar software and sell it at a lower price to smaller government agencies that could never afford Palantir—or grasp its software. He was right, and at the moment, Peregrine has been averaging a half-dozen new contracts a week. The typical contract brings in about $250,000 in annual revenue, though some are for as little as $30,000, and usually the deals last three to five years.
Because the software needs to pull in data from all sorts of places, it’s complicated to set up. After a customer signs a deal, a Peregrine employee spends much of the next three months troubleshooting onsite with staff. Those employees generally have a data and software background. Peregrine also employs a number of former cops, who will travel and help their tech-minded colleagues understand what police department customers actually want.
Peregrine’s browser-based software, which relies heavily on AI, can parse both structured data (such as spreadsheets and databases) and unstructured data (images, PDFs, video). It can, for instance, create a 3-D map of a wildfire’s spread that shows where various models suggest it will go next. It can show whether schools are within the blaze’s path and provide a real-time display of ambulances and police cars in the area.
As I mentioned, the software has a powerful search function, too. If an officer searches for “swastika,” say, they will find every mention of the word within the transcripts, records and files uploaded and connected to Peregrine. A search for “swastika tattoo” will turn up both every written instance of the phrase—as well as every image the department has of a swastika tattoo, with the AI identifying the symbol and the fact that it’s a tattoo. Say an officer is looking for someone with a swastika tattoo on their left ankle: A search for “swastika tattoo left ankle” will quickly display every photo the department has of a left ankle with a swastika tattoo on it.
Peregrine’s software resembles the technology I picture in a Hollywood espionage thriller, the kind that would seem too simplistic to exist in real life. So when Noone showed it off to me last week at Peregrine’s Manhattan offices, I gawked at it. And then it gave me a profound sense of the chills. Watching Noone reveal Peregrine’s neat tricks was the same eye-widening experience I’ve gotten from uncovering a new use of ChatGPT—except, of course, Peregrine deals in matters of life and death.
Noone insists Peregrine greatly limits any chance of police misuse because it allows agencies to restrict what data each user can access. A patrol officer, for instance, might be able to see only a fraction of what a homicide detective can. Still, it’s pretty easy to imagine someone accidentally receiving more access than they should have.
Another thought then crossed my mind. If Noone, a person professing his good intentions, could create Peregrine—a much more powerful police AI than I could’ve imagined—then it’s chilling to consider what a person with malicious intentions could assemble, given the tremendous advancements in the underlying AI technology behind Peregrine.
It made me think pretty instantly about “Minority Report,” Steven Spielberg’s cautionary tale of police analytics and efforts to predict crime. I asked Noone whether he could easily make a Peregrine feature that tries to predict someone’s habits or whereabouts based on information fed to it by the police.
“With any one company is the capacity to do a lot of things: Of course we could,” he said. “Do we? No.”
Since Peregrine makes software the public never sees, it has largely escaped backlash from citizens and privacy advocates. The same can’t be said for Flock Safety, an Atlanta-based crime tech startup that has been a consistent magnet for ill feelings.
Flock’s main product is a license-plate–reading camera that it sells to cities, mounting the cameras on poles in public areas such as busy intersections. Since they’re relatively inexpensive—just about $3,000 a year—cities install them en masse. Sometimes Flock sells them to private businesses and groups like homeowners associations that want to monitor passing cars, and sometimes those organizations share their footage with the police.
The cities’ police departments can access the footage through Flock software. Police can use the program to find an image of a specific license plate and set it to alert them in real time when it spots that plate. Officers can also use natural-language prompts within an AI-powered search: For example, “blue F-150 truck” will find footage of blue F-150s. Another AI feature promises to identify possible accomplices in a crime by highlighting vehicles spotted near one police are already tracking.
Just in the past week or so, Flock cameras have attracted negative headlines everywhere from Greenville, Miss., to Cambridge, Mass., to Santa Barbara, Calif., with critics mostly blasting them as an invasion of privacy.
The ACLU’s Jay Stanley thinks the Flock cameras will usher in a period in the U.S. where “anybody living in this country could find themselves in the crosshairs of law enforcement just because some algorithm—about which we know nothing—has decided that your driving patterns are suspicious.”
Flock Safety CEO and founder Garrett Langley, 38, thinks opponents to his technology “enjoy a privilege that I don’t have: which is they get to debate hypotheticals and present no solutions to a known problem”—namely, crime. And he feels the ultimate responsibility for how his tech gets used should fall squarely on local governments.
“That puts the onus on our elected officials,” he said. “That’s why we elected them.”