Crunch data from your car for fun and profit

A $99 device and a Spark back end creates an ecosystem of car-connected data and applications

tunnel light speed car

Automatic is to your car what Fitbit is to, well, you. This $99 device plugs into your car's diagnostic port (OBD-II), which, on newer cars, is located by the fusebox -- and connects to any iPhone or Android device running the Automatic app. It captures highly useful driving and diagnostic data, and it even sends out a distress signal if you're in a crash.

The tool lets you find out to what degree the EPA sticker on the side of your car that claims 38 MPG is lying. It also analyzes your driving patterns, makes suggestions, and tells you when you need maintenance.

The company has its own analytics, of course, but it also has a REST API in alpha -- and is working to create a third-party ecosystem. For instance, you can use your Automatic data to record your mileage and directly download it to Quickbooks if you use your personal car for business. Rob Ferguson, the director of engineering at Automatic, told me this could be especially important for Uber or Lyft drivers keeping track of which miles they drove for which company.

Automatic car app

The Automatic app records your driving habits and makes recommendations to raise your MPG.

I'm rather dubious of the surveillance state, except when it comes to my kids. They never have to give me their passwords or anything -- because I have Wireshark, root on the Wi-Fi router, and rooted their devices. (Now you know how my youth was misspent.) Anyhow, Automatic can help you extend your parental PRISM and monitor teen driving:

     "id": "E_63db5c25ffd955ba",
     "user": {
      "id": "2fcd8161801e95d1e615"
      "type": "notification:hard_brake",
      "created_at": 1383448450201,
       "time_zone": "America/Los_Angeles",
       "location": {
        "lat": 37.757076,
        "lon": -122.448120,
        "accuracy_m": 10
      "vehicle": {
        // Vehicle Object
     "g_force": 0.3

The APIs have a number of implications. You can integrate with your Jawbone Up, so when you drive to a location less than two miles away, you'll be nagged to walk next time (and told what you could have saved on gas). Moreover, you can have your car integrate with your Nest device using the Automatic ETA APIs and autoset your thermostat; that way, when you walk in, it's a perfect 65 degrees Fahrenheit (which my wife sets back up to what feels like 90).

“We really are trying to embrace this 'Internet of things and quantitated self,'" says Ferguson. "Think about all of the ways your car is part of your daily life and how would you want to interact with it.” In fact, if you sign up with Liberty Mutual Insurance, they will buy the device and give a discount of up to 30 percent on your car insurance if you are a good driver -- which they will know because you've opted into the monitoring. They'll show you the factors that go into the evaluation. This reminded me of an ACLU video where your pizza joint was integrated with your health insurance provider.

I asked Ferguson whether I really wanted a total Internet-of-things-enabled surveillance network run by corporate America in addition to the one run by the NSA. “Truthfully, the way I think about it is that these people are going to get to your data, and it is a lot better if you can opt in and say, 'Hey, I want to give this data because I'm a good driver,' or be able to opt out and know your data is yours. We're the harbinger of it.”

Automatic is using Spark and machine learning along with Cassandra, DynamoDB, and a slew of other new data architecture technologies. According to Ferguson, Automatic “really believe[s] in the microservices architecture -- lots of small services that do specialized tasks. If one thing needs a memcache and one thing needs a Redis, we use exactly the right technology for that service.” Yet because much of the data is relational, Automatic uses a lot of PostgreSQL, too. Ferguson also noted that Spark was a “game changer” for the company. For machine learning and interesting analytics, Automatic does everything with Spark.

The ODB-II standard doesn't detect what kind of gas the car is using, so Automatic used analytics based on the ideal gas law to figure it out. Moreover, by hooking into GPS, they can tell which gas station you went to and how much the gas cost.

The company is also trying to use its analytics for the greater good and identify dangerous intersections before they become dangerous. Ferguson says, “It's really expensive for municipalities to find out how much traffic is moving through their city and what are the actual costs and dangers involved.”

Automatic is also starting to look at drag and load based on the ideal physical model of your car. Based on this, they can make recommendations and tell you how much it costs to keep your bowling ball in your car. You can also compare your Civic to other Civics and see how you’re doing in MPG compared to others.

“[Before working at Automatic], I didn't necessarily know a lot about the different factors that affect fuel economy," admits Ferguson. "For instance, in winter, when you make all of those short trips and blast the heater, you waste an insane amount of money. You'd be better off leaving your car running the whole time than making those short trips.”

Ferguson said all the things companies usually say about user data security and how careful they’re being (right before they publish a creepy blog on one-night stands) but noted, “The way that we’re going to make our next generation of data products is by doing analysis on all of this transportation data that we’ve pulled in.”

So I bought one. I’m looking forward to doing some strange analytics.

Copyright © 2015 IDG Communications, Inc.