Americans are finicky about their cars. We treat them more like appendages than machines–they symbolize personality, independence, and sex appeal all at once. Yet as part of a culture of car lovers, most of us know embarrassingly little about what goes on under that 5.9% APR-financed hood. In the age of Big Data, ubiquitous computing, and personal analytics, why do we still know so little about what’s happening inside our cars?
Automatic, a personal-driving-assistant app designed by a team of Y-Combinator-backed Berkeley School of Information grads, wants to teach you. “The amount people spend on their cars is high, but the amount they know about their cars is low,” says Ljuba Miljkovic, Automatic’s chief product officer. “Small changes in behavior can lead to large savings.” Automatic does four things: It keeps track of where you park, notifies 911 in the case of a crash, monitors engine performance, and most important, teaches you how to drive more efficiently using a series of subtle audio cues. The app leverages information from your car’s onboard computer to make you a better driver–not just in terms of safety but in terms of fuel efficiency. Think of it as hypermiling for the average consumer.
On a rainy afternoon last week, Miljkovic and Automatic CEO Thejo Kote stopped by to take me for a joyride in their Automatic-enabled Zipcar. We head down to the garage, and when their silver sedan coughs to life, a small bit of white plastic beneath the steering wheel glows in acknowledgement. This is the Automatic Link, the single piece of hardware required to run Automatic. The Link acts as a translator between your car and the Automatic app running on your phone. It plugs into the Onboard Diagnostics Port, a small outlet standard on every car manufactured since 1996. Before now, I hadn’t even realized the ODP existed–apparently, that’s normal. “Your car is the most expensive computer you own,” Kote says. “Yet it’s a black box. We wanted to change that.”
As soon as the car starts, Automatic begins a new entry on the app, which is running on Miljkovic’s smartphone. It records everything from velocity to RPM and fuel efficiency, but it synthesizes all that raw data into a few simple metrics, like how much cash you spent on gas for a particular trip, a map, and a log of significant events. The most important stat, though, is your “driving score,” a number between 1 and 100 that reflects how efficiently you’re driving. Speed up or brake too quickly, and your score goes down. Same for speeding in general, which burns outsized amounts of fuel. As we wheel out of the garage, Miljkovic points out that the app doesn’t show you any stats until you’re back in park–a necessary safety measure, lest everyone drive around with their eyes glued to their phones (well, more than they already do). Instead, Automatic communicates with the driver through a series of subtle audio cues, a model of behavior modification at the core of Automatic’s user experience.
At Berkeley, Kote and his partners were deeply influenced by B.J. Fogg, the Stanford researcher whose seminal thesis, Charismatic Computers, investigates how technology can affect behavioral change. Automatic is built loosely around Fogg’s model, which posits that we change our behaviors when confronted with three necessary factors: motivation, ability, and trigger.
With Automatic, the “motivation” is savings and safety; the “ability” is the driver’s ability to alter their habits on the road. The trigger is the tricky part–it needed to be simple and clear but not so obvious that it distracted the driver. Miljkovic engineered three subtle chime noises, one for each bad habit (hard acceleration, hard braking, and speeding).
For the first 10 minutes of our drive around lower Manhattan, I hear nothing. But speeding up the West Side Highway, we break hard at a stoplight, and a metallic ding emits from the device. It’s about the same intensity as the audio prompt played by Apple’s Mail app. If I hadn’t been listening, it would’ve faded into the background of our conversation, unnoticed. The audio design is the most subtle element of the app, but it’s also the most important: Fogg argues that successful behavioral modification happens when a technology delivers the right cue at the right time.
Kote and Miljkovic are excited about the idea of making drivers “smarter,” right down to what your car’s dozens of engine light notifications mean. To demonstrate, they’ve unplugged the sensor that monitors the engine’s gas intake–an orange “check engine” light blinks on the dash, and on the smartphone a notification appears. I tap it, and it opens a dialogue box showing (in plain English) what that particular code means. “We license the database and do some of the copyediting ourselves to make them easier to read,” says Miljkovic. Yelp suggests the nearest mechanic, but you can also choose to turn the light off if it’s not serious.
After a spin around the city, we park and wait a few seconds as the app finishes processing the data. Then, a new entry appears alongside a map. According to the app, we spent about a dollar on gas during our 20-minute drive. We braked too quickly twice. And our overall driving score, based on safety and savings factors, was a respectable 96 out of 100 (they avoided a letter grade system, for obvious reasons). Based on data from the EPA, a driver who keeps their score in the 90s stands to save up to 35% on gas. Consider that the average driver spends $3,000 a year on fuel. Then consider that there are 250 million cars on the road today. Automatic’s $70 price tag begins to seem like chump change.
Like Fuelband for fitness or Nest for thermostats, Automatic reads human behavior and nudges users toward better habits. “Driving isn’t just about getting from point A to point B,” says Kote. “Owning a car is an experience. We think there’s a lot of room to improve that experience, in ways that no one is focusing on.” The next step is to scale up production, which’ll give the design team a chance to calibrate their thinking on how the app should evolve as the company grows. Eventually, that could mean negotiations with automakers, or allowing users to dig into their data on their own. For now, Automatic is on pre-order, and the team expects to ship in May.