In June 2016, Antoine Balaresque, the cofounder and CEO of the hot new startup Lily Robotics, stood before a room of business students at Berkeley’s Haas School of Business, ready to reveal the PowerPoint slides that had made him an instant startup celebrity. Wearing the ubiquitous Silicon Valley uniform of a T-shirt and jeans, he appeared slightly bashful, with unruly hair and a boyish face still round in the cheeks. He seemed self-conscious about being feted by the room of business school students.
Jessica Pishko is a San Francisco-based journalist who writes frequently about incarceration and social justice issues.
Sign up to get Backchannel’s weekly newsletter.
The presentation began like most of Balaresque’s talks, with the Lily Drone promotional video: A slick film showed the drone swooping through the air, capturing footage of users engaged in a series of outdoor adventures. When the video finished, Balaresque began to recount the origin story of his “flying camera.” It started in 2013, with a family trip to Yosemite National Park, during which Balaresque’s mother took a group photo. His mother, he said, was always behind the camera, so “she was missing all these memories.” His mother’s absence from the shot inspired him to create a selfie drone—the Lily Drone—that would be portable and easy for novices, like his mother, to use.
There are many camera drones on the market, but along with cofounder Henry Bradlow, Balaresque had created a product with a unique attribute. “It flies itself,” Balaresque told the Haas crowd. Using a combination of a GPS tracking system and visual recognition, the pair designed the camera drone to follow users wherever they went—like magic, without the need for a remote control. It was lightweight and portable, designed for both the novice traveler and the hardcore adventurer.
Balaresque was solidly at the helm of his story, and it seemed like his company was going stratospheric. The previous year, Lily Drone had enchanted Silicon Valley and beyond. In 2016, the Wall Street Journal put it on its list of products “that will change your life.” Balaresque and Bradlow were named in Fortune’s 30 Under 30. Facebook buzzed with excitement, and people eagerly placed $499 preorders, imagining the drone on family trips and skiing adventures.
Just a few months after this presentation, by January 2017, the headlines had changed. “Drone Startup Abruptly Shuts Down.” “Is Lily Robotics the Theranos of the Drone World?” Lily Drones was now “hyped,” “collapsing,” and “failed.” Preorder customers bemoaned their losses on the internet. Those who hadn’t ordered gloated. In early 2017, the company declared bankruptcy and was sued by the San Francisco District Attorney’s office for false advertising based on claims that the promotional video—the same one that elicited applause less than a year ago—was fake. The DA’s office accused the founders of publicizing a product that they knew wasn’t possible to make in the timeframe advertised.
Over 60,000 Lily customers are still waiting for their drones. But was the Lily Drone, as the headlines suggested, all a con? The trajectory of Lily Robotics is a cautionary tale for the young and adventurous enchanted by the latest technology dreams. Though 3D printers have revolutionized at-home manufacturing, there’s still something profoundly difficult about rolling out a phalanx of sleek flying drones without experience and expertise. The story of Lily is about two ambitious college students with smarts and personality who wanted to change the world—or at least photography. But they didn’t have the right tools, and didn’t listen to those who did.
The story of Lily Drones starts at the University of California, Berkeley, where Balaresque and Bradlow were students. As a teenager in France, Balaresque was inspired to apply to UC Berkeley by the cousin of a classmate who attended. At Berkeley he was first exposed to robotics and science—an interest that solidified when, as a Business Administration major, he met his cofounder Henry Bradlow, another undergrad studying computer science.
The duo arrived at Berkeley at a precipitous time. Where Stanford University had dominated the Palo Alto tech scene since there was one, Berkeley had largely failed to produce the kind of high-profile startup wunderkinds flooding the Bay Area. When the pair arrived on campus in 2010, the university was ramping up its investments in seed funds and student-run startup competitions in an attempt to generate more entrepreneurial drive on the free-spirited campus.
Lily benefitted from “Demo Day,” a startup competition at Berkeley, where Bradlow and Balaresque received $400,000 in seed money from angel investors. By March 2014 the pair had joined Skydeck, a UC Berkeley-based accelerator with a six-month “growth and survival” program. Just a month later, Lily received $1 million in seed money from the Dorm Room Fund, a student-run venture capital firm started by Jeremy Fiance, an undergraduate classmate of Balaresque and Bradlow. The fund was created to aid Berkeley students who had startup ideas but no capital. (The fund is now backed by First Round Capital, which receives a cut if any student projects work out.)
Because recipients of Dorm Room Fund grants are students, they aren’t subject to the same expectations as more established entrepreneurs. Rei Wang, the current director of the Dorm Room Fund at First Round, told me that it’s “hard for people to be a full-time student and a full-time funder.” She continued: “A few users have a working prototype, but not much beyond that…[we look for] a big vision and commitment to the idea.” Though Wang was not directly involved with the fund’s investment in Lily Robotics, she added that Balaresque and Barlow were “consumed with solving a problem [and] passionate about storytelling,” even if, like many undergraduate entrepreneurs, they were “less mature.”
Yet thanks to the school’s investments, the two had enough money to hire a real hardware architect who could help them build working prototypes. They moved from a closet-sized office in Berkeley to a garage space with technical equipment and 3D printers in Atherton, close to Silicon Valley. And one of their first tasks was to use their new funds to create a kickass promotional video that would help them raise money through presales.
The Lily Drone promotional video begins in motion. A downhill skier tosses the device into the air like a Frisbee and then skis away with the drone camera trailing. The drone itself looks like a sleek Apple product, with four spokes emerging from a central sphere and a GPS guidance system that the user wears around the wrist. The video’s emphasis is on portability: People shove the drone into backpacks; they drop it into water; they launch it, again and again, into the air. At one point a snowboarder throws it off of a bridge, and it returns like a boomerang. (According to a DA’s office interview with the producer of the video, one prototype didn’t come back.)
If Balaresque and Bradlow—young men raised on the iPhone—knew anything, it was that their product needed to be cool. In early 2015, Balaresque and Bradlow hired Brad Kremer, a director well known for his radical snowboarding shots. According to interviews conducted by the San Francisco DA’s office, Chris Frey with CMI Productions, a production company focused on tech companies, said that he was hired along with Kremer to help Lily Robotics produce what would be its launch video. Balaresque and Bradlow came to Kremer, he said, with a problem. Lily had spent the bulk of its money on another director, who had created a video with which they were unhappy. The video needed to be all about the “product promise.” So the pressure was on to succeed.
The video was shot on location in Tahoe in February and early March of 2015. According to Frey, Balaresque and Bradlow shot some video with the Lily Drone and some with a DJI Inspire—an expensive competitor product commonly used in feature films like Lord of the Rings—which a user controls remotely. In an interview with the DA’s office, Kremer said that the company “trained him” to mimic the Lily drone camera with the Inspire. Because the Inspire featured a rotating camera that could shoot from more angles than the Lily, Balaresque and Bradlow told Kremer to make sure that the Inspire maintained stiff shots—more like the footage a Lily camera would make.
According to his interview with the DA’s office, Kremer sent the Lily team a rough edit of footage from the three day shoot. The usual process, Kremer said, would be for the Lily team to make suggestions. But shortly after receiving the cut, Balaresque emailed Frey and Kremer, asking for the raw footage so that he and Bradlow could make their own edit. This second, cobbled-together version became the final version, released to the public in May 2015 .
Only Lily’s founders, who cut the final video, could definitively say how much footage came from a Lily Drone prototype, both filmmakers told the DA. In the video’s closing moment, a family matriarch throws the Lily Drone into the air to catch a shot of her family and friends, smiling in a field. According to Frey, that scene was shot with both a Lily Drone and an Inspire—making it impossible to know if crisper Inspire footage had been substituted. (In his interview, Kremer said that the shot had been captured using the Inspire, not the Lily. He explained to the DA’s office that Bradlow and Balaresque asked him to shoot the scene with the Inspire, just as a backup, simply because they were losing the light.)
“In terms of the final edit, I’m not sure there’s any Lily footage,” Frey told the DA . “I’m not 100 percent sure.” This uncertainty troubled Frey enough to remove the video from his company’s website after Lily was sued.
Some of the features the video touted weren’t functioning as advertised. In his interview with the DA, Frey said Lily’s “follow” feature, which allowed the drone to trail a user and film, was working. But according to Frey, the team was still working on the software for the other features advertised in the video. Frey told the DA that the first “throw and go shots”—scenes where the user throws the drone up in the air—were “a disaster.” “It did not work on any sort of reliable basis,” he added. Balaresque and Bradlow scrambled to fix their prototypes on the mountaintop, suspecting that it was the cold and the altitude, from their filming location at the top of Tahoe’s Mount Rose, that was causing the bugs.
An early technical employee, who—like other Lily employees—spoke to me on condition of anonymity because he signed a nondisclosure agreement, told me that, though there were multiple working prototypes of the Lily drone, the “camera hardware wasn’t far enough along to do high quality shots.” He emphasized that the initial Lily drone models were built with ready-made parts that weren’t customized for the special features the team had envisioned. Mostly, the source said, the color was off and some shots were blurry. Yet several ex-employees said that Frey’s report was exaggerated, and most of the drone’s functions were performing as advertised when the video was filmed.
As soon as the promotion video came out on May 12, 2015, people began to purchase the drones. Quickly, the video amassed 30 million viewers, and Lily sold $34 million in presales at a discounted price of $499 each. (The price increased steadily in $100 increments as the presale period ended, up to $999, the intended retail price.)
In 2015, based on the outpouring of public interest, the company received $14 million in funding from sources including Spark Capital, The House Fund (Jeremy Fiance’s post-Berkeley startup fund) and Winklevoss Capital.
There were signs, of course, that the Lily Drone wasn’t as magical as it could be. In one video, released May 12, 2015, a reporter with The Guardian went with Balaresque to Sheep Meadow in Central Park to test the drone. They tossed it into the air, and it dropped like a stone. When it finally took off, it veered away from the user and crash-landed on some sunbathers. Balaresque pulled out a screwdriver and did some quick repairs. Finally, it hovered anxiously and, according to the review, took a handful of photographs.
I tried multiple times to contact Balaresque and Bradlow both through social media, email, and acquaintances for comment. I received no response.
Several employees involved with Lily Robotics thought the legal proceedings involving the video were overblown. They described the video as an advertisement: It depicted the best version of reality, not reality itself. One college friend of Balaresque and Bradlow did not think that there was anything unusual about the video: “Of course it wasn’t misleading; it’s not a thing yet. That’s why you are doing a prelaunch.” Someone close to the company compared it to a car commercial, where the vehicle does dramatic swerves that no one would test in real life.
Lily promised delivery in early 2016. Its preorder form was incredibly simple. There wasn’t even a spot to put your address—it was just name, email and credit card information.
Then people waited for their drones to arrive.
Before the preorder period closed on October 6, 2016, Lily had descended into chaos. The $14 million wasn’t enough to cover the costs of hardware construction, which, according to employees, Lily had outsourced to a Chinese contract manufacturer. The internal struggle, according to two former employees, was both developmental and managerial. After the spectacular presale, Lily began to hire over a dozen people, not including a batch of interns. They leased a hip San Francisco office. “The more money they got, the less they listened to advisors, and the more money they spent,” a former employee told me. “They got millions of dollars and [thought], ‘We are special.’”
There were production hiccups. An engineer who led the software development team insisted on revamping the drone software to be his own original invention, several engineers told me. (The prototype had been made with open-source software.) The engineering team rebooted and the drone prototypes stopped flying. Production was set back about six months.
In the fall of 2015, Lily took out a $4 million loan using the pre-sale money as collateral. People on the internet began to get anxious. Still, things seemed to be going well. In November 2015, Lily announced via the company blog that units were rolling off the production line. It also won “Most Innovative Product” for the 2016 Consumer Electronics Show, around the same time that the Wall Street Journal heralded it as an above-the-line success.
Then the bubble burst. As early as December 2015 , Lily sent emails to customers alerting them that production would be delayed by several months, all while assuring consumers that it was not using the preorder money to pay the bills. (A source confirmed that the company never did spend the preorder money except to reimburse people.) Instead, Lily said that it had a new round of private investments from Spark Capital and other entities, which, according to the company, was designed to inject some cash into Lily to make it more viable for purchase.
Then, in August 2016, Lily issued notification of another delay, pushing the timeline out to sometime in 2017. The company sought a buyer for an infusion of cash that would allow it to make enough models to meet consumer demand. Rumors circulated that Snapchat was exploring a purchase, and that it then backed away. In December 2016, Bijan Sabet, from Spark Capital, left Lily’s board. Some employees were laid off with no warning. Competitor products began to pop up. In the fall of 2016, at least one employee left for a drone company founded by ex-Google employees that was producing a very similar product.
In early 2017, the San Francisco District Attorney’s office raided Lily’s headquarters, carting away an assortment of drives and equipment that it has not returned. It also filed a motion enjoining Lily from spending any more money before it refunded consumers. The SFDA also filed civil consumer protections charges against Lily, relying on the video as false advertising. Both cases are ongoing.
The SFDA’s office used emails between Balaresque and the filmmaker as evidence in an attempt to show that Balaresque intended to deceive the audience into purchasing the drone. Early on in the filming process, Balaresque emailed Brad Kremer at CMI Productions, the company Lily Robotics hired to produce the video, about his concerns that a “lens geek” would study the film up close and see that the shots were clearly from an Inspire. “But I am just speculating here,” he wrote. “I don’t know much about lenses but I think we should be extremely careful if we decide to lie publicly.”
Ex-employees disagree on where Lily went wrong. I was repeatedly told that hardware (like a drone) is much more difficult to fund and produce than software (like an app). Other sources told me that it wasn’t the hardware at all—it was a failure of management. “It was a combination of hubris plus optimism,” someone close to the company said. Balaresque and Bradlow were not polished enough to be able to secure further funding. “I felt sorry for them,” that source said. Multiple ex-employees described Balaresque and Bradlow as hard workers who underestimated how capital-intensive drone production was.
What does it mean to want something that doesn’t exist yet? The magic of the Lily Drone was in its concept: It was a product you could unpack and throw—so easy, Balaresque wrote in emails, that even an old person could do it. But translating that idea into a tangible product proved difficult, and the storytelling that made the Lily Drone so tantalizing to consumers ultimately factored into its downfall.
In one of his presentations, Balaresque presented a PowerPoint slide with the sentence, “Humans have a fundamental need to put themselves in the center of stories.” It appeared to be a quote he made up, but the idea that human nature needs stories is fundamental. Stories are how we make sense of our lives. But while a good story can get you funding and acclaim, ultimately it isn’t enough.
To figure out what was special about the promise of the Lily drone, I went back to that final scene, where an extended family stands in a beautiful field—Ireland? Scotland? Marin?—and the grandmother throws the drone into the air. It circles about, and everyone jumps and waves at the camera. The camera soars higher, focusing first on the scenery, until it swoops back into the people, all with looks of joy and exertion on their faces. The family presumably knows nothing about robotics: It just wants that moment preserved forever.