When Google announced Android Auto at Google I/O 2014, I was already sold. And by "sold," I mean I fully expected it to be something I'd want [were I in the market to buy a car that had it]. And while I don't actually plan on buying a car with Auto any time soon, after spending a week with it, I do feel pretty OK with that gut feeling. We reviewed Auto earlier this month on a Pioneer head unit, but I figured I'd also share my own thoughts on it.

For a little bit of background, recently Hyundai allowed me to borrow a Sonata sedan (I reviewed it) with Android Auto loaded up. If you have a 2015 Sonata with the tech package (the 8" nav system, basically), you should be able to get the Android Auto upgrade at your local dealership by now. So, hundreds if not thousands of people are already driving around with this, experiencing it on their day-to-day commutes.

Hyundai's version of Auto does have at least one perk over a third-party head unit - Hyundai claims they use the vehicle's GPS instead of the phone's for navigation, for example, and that they've fine-tuned the experience and performance to an extent that should make Auto robust and difficult to badly break for drivers. This is important, too: safety really should be the number one concern of anyone getting behind the wheel, and as a mass-market automaker, Hyundai is obviously going to try to ensure that their products don't cause unnecessary driver distraction.

My first day with Auto, I spent about an hour in the car and used an ASUS ZenFone 2 as my connected device. This was a terrible idea. You see, at the moment, Android Auto seems rather specifically optimized for certain devices. On the ZenFone 2, the interface chugged and lagged so bad that even a car manufacturer would probably be embarrassed by it, let alone a software company like Google. The phone also lost battery capacity while attached to the Hyundai's USB port even though charge was flowing. Over a 30-minute drive using navigation and streaming Spotify, the ZenFone 2 lost 4% battery. On the slightly shorter return drive later that day, it lost 2%.

So, on day two I tossed the whole ZenFone thing - it was clearly providing me a bad example of the typical UX. I don't know if that's down to Intel or ASUS or Google, but suffice it to say I would not even want to use Auto if this is what it was like on every device. Instead, I plugged in a Galaxy S6 Edge. The S6 Edge performed far better, with the interface moving along at a fairly consistent 25-30FPS (that's an eyestimate) without too many unexpected delays or pauses. The phone also actually charged, gaining 2% battery in around 20-30 minutes of driving, which while not particularly good, is much better than losing battery capacity. But then, the next time I drove, it did lose capacity. A lot... because I was driving in an area with poor data coverage. Cell reception seems absolutely critical to Auto's power usage, because when I wasn't able to get reliable LTE for music streaming and was forced onto 3G, the battery dropped substantially more.

Anyway, it's from here that I'll recount what using Auto is actually like, what I enjoyed about it, and what I didn't. My first day with Auto was bad enough that I don't think I'd be helping anyone by saying I was ready to physically throw my phone against a wall for all the grief it was causing me. On the S6 Edge, things went much more smoothly. Here are my thoughts.

The best parts are generally very good.

Your initial plug-in with Auto is pretty seamless on the Sonata. Just download the Auto app on your phone, attach your phone via USB to the car's port in the center console, and then tap "Android Auto" when the icon appears in the homescreen of the Sonata's infotainment UI. You'll get a disclaimer about distracted driving and such which you can set not to appear again (yes), and then you get an option to dive into a tutorial.

As you're probably aware, there are essentially two ways to interact with Auto - voice and touch. I chose to do pretty much everything by voice by the second day, because it was simply more natural and less distracting while I was actually driving. Playing with an embedded touchscreen is, to me, no less dangerous than tapping at a phone attached to your dashboard when you're trying to focus on not crashing into people or objects around you. Activating voice commands happens with a long-press of the voice button on your steering wheel controls, and you get the same "bloop" prompt as you do when using voice actions on Android to let you know when to start talking.

DSC08594

Voice input in Auto is prompted by long-pressing the voice command button on the Sonata's steering wheel

This is essentially the crown jewel of Android Auto's functionality right now - voice recognition that doesn't suck. Using the car's built-in microphone, Google's voice actions really come into their own when you're driving. Want to know how long it's going to take to get home? Just ask Google. Literally, just ask that: "How long will it take to get home?" Boom: you get a navigation preview with estimated time and traffic levels. Yes, this could be done with any car with navigation, adequate scripting, and live traffic information, but Google was literally built to answer these kinds of questions. Context-awareness and natural language processing give Google's voice actions a massive leg up over the built-in systems on most vehicles because Google effectively understands that you can ask for the same answer fifty different ways, and Google is constantly trying to get search to understand many more. Let me provide a few examples of voice actions really working well in a car.

  • "Where's the nearest Chevron gas station?"
  • "Play classic jazz on Spotify."
  • "Navigate to the nearest post office."
  • "Send a message to Dan on Hangouts: I'm running about 15 minutes late."
  • "Find sushi restaurants within 5 miles."

Basically, you can do the stuff you should be able to with voice commands in a car now... which is admittedly still a pretty short list of stuff. Android Auto also isn't brilliant all the time, though. For example, if I say "Play Tom Waits on Spotify," it knows what I want. If I say "Tom Waits on Spotify," it doesn't. If I'm in the Spotify app's UI in Auto, Auto will try to open Google Play Music if I give a voice command to play a specific track or album without specifying an app, and I don't use Google Play Music, which leads to a dead end. That's exceptionally annoying. The system should absolutely be aware of current context, that I have to say "on Spotify" while currently playing music in Spotify with the Spotify app displaying on my screen is clearly a bit silly.

DSC08693

As far as mapping, for finding basic points of interest like ATMs and gas stations, Google Maps is just great - there is nothing better. And I want to really give Maps praise for that, because it deserves it - Google tirelessly works to remain the leader in mapping software and it always shows.

Still, in terms of discovery, if you're browsing for a good restaurant or bar in an unfamiliar city, you're probably going to want to stop somewhere and pull out your phone and use another app. And this is because I'm guessing most of us have experienced the whole "Google Maps is a review wasteland" syndrome outside of New York and the SF Bay Area. Because Auto has no alternative POI or mapping clients at this time, it's Maps or nothing.

And if you know where you want to go, Google Maps is great. It's the best, even. If you don't, well, the ecosystem just isn't there yet for alternatives. Not that I'm saying this is terribly surprising, it's just one of those things that makes you realize that Auto is not a full "in-car smartphone replacement."

SMS and messaging leave something to be desired.

What about message handling? Isn't that supposed to be one of those things we all want in the car - easier, safer ways to manage text messages? Well, for now, I find that on the receiving end, Android Auto is no better at this than pretty much any of the existing enhanced Bluetooth technologies available on many vehicles. It reads incoming messages aloud, and you only get the first 5-10 words of a given message in the Auto UI - there's no way to expand messages and see more, let alone the full message. They have to be read back to you.

DSC08661

Note there is no dedicated message UI - and that's for a reason.

This is plainly a distracted driving consideration, but at the same time, why on earth shouldn't I be able to quickly read a Hangouts message or SMS without disrupting my music when I'm at a complete stop? If anything, focusing on the hilariously slow and awkward reading of the Google TTS engine distracts me more because I have to listen to. every. single. word. closely. Either the TTS engine needs to get a lot smarter or Auto needs a way to give me my message content when the car is not in motion. Google discourages the latter by putting a big, blank "Android Auto" splash on your smartphone when it's running Auto, but getting around it is super simple - just tap/wake the phone a few times and it usually goes away and stays away.

Once people figure this out, they're just going to start pulling out their phones again to check texts again - making bad behavior slightly less convenient won't stop those people most likely to engage in it anyway. And if a text message is considered too "dense" for safe reading on a vehicle, let me introduce you to, I don't know, literally every car manufacturer's byzantine settings and infotainment UIs that are largely accessible for viewing at 70MPH. If those are "safe" to read (even if some functions are locked out), I'm not sure what the point in pretending a 15-word SMS is comparatively deadly is. It's rejecting practical reality for principle. I can understand not wanting to let people read texts of children's-short-story length, but the current setup is restrictive to the point of annoyance. As long as you're not giving people a keyboard and limiting the maximum message length to, say, tweet-sized, I don't think there's a massive safety concern here.

Sending SMSs in Auto is a much better experience, thankfully, and that's once again because of functional speech recognition (as compared to that in use by most automakers). I don't generally even worry about it screwing up the content of my messages because I really find it that reliable. Considering how bad speech recognition can be in some cars, this is a quantum leap in usability. It's also a huge plus in terms of safety and general driver attentiveness, because typing on your phone while driving a moving car is essentially death-wish grade stupidity.

The big question: how ambitious can Auto realistically be?

So, aside from music, mapping, phone calls, and texts, what does Auto do, exactly? For now, not much. You can see your Google Now weather cards, your reminders in Google Now, and... that's pretty much it. I feel like there's a need to dispel expectations that Auto is simply your Android phone projected into a pretty, car-friendly interface. It's decidedly not, and it shouldn't be. Most of your notifications won't even show up because Auto only allows access for Auto-enabled applications, meaning those applications must explicitly be approved by Google and follow the Auto app guidelines for safety and functionality. You'll get calls and SMSs for basic telephony functions, but the only other content can come from Auto-enabled apps (of which there are currently sixteen not made by Google). As for hacks and workarounds to get "custom" or unapproved Auto apps on your car? When speaking with two Google reps, it was suggested to me this was unlikely - there are allegedly mechanisms in place to prevent exactly this kind of tinkering.

Once again, it's clearly a safety thing: you should not be checking your email in your car. This is common sense. You should also not be on reddit, managing your Tasker profiles, or picking out a new calendar app. You should do these things on your phone, while relatively stationary, and preferably not at the wheel of a vehicle. This, I think, is the bucket of cold water that will make the tinkering and customization community mildly despise Android Auto. There is no making it your own, there is no flexibility, and there seem to be rather strict ground rules about what is and is not "acceptable" behavior in the car.

In part, this is because Android Auto is designed to be ludicrously mainstream and safe. Google wants a simple, straightforward experience that the average person of even middling motivation is going to be able to figure out without too much trouble and, perhaps more importantly, will be tremendously unlikely to be distracted by. Android Auto's interfaces are so sparse that sometimes I question the use of screen space. Google's directive one seems to be telling developers to display as little text as humanly possible, and in some situations that's fine, but in others I'd like just a twinge more information.

DSC08662

The black bar on the bottom here as well as the white gradient at the top seem like unnecessary obstructions in a map view.

Given that Hyundai is the only vehicle manufacturer shipping Auto yet, there's also no real comparing how Auto will work in other cars. There's a whole "5th tab" dedicated to vehicle metrics and similar information in Auto (the gauge icon on the far right), but it doesn't do anything on the Sonata, and I've not heard that Hyundai has any plans for it at this time. It seems to be a "blank" reserved for carmakers who will eventually release an Auto-enabled smartphone app or allow Auto to interface with the vehicle in other ways. How exactly it will do this, I have no earthly idea. While it's unlikely such apps will do anything the vehicle's native interface could not (and would likely do substantially less overall), this would be a plus in terms of not having to exit Auto to get some of your vehicle's native functions and features. But this is speculation, we don't know what this tab will do or if carmakers will even use it. Google hasn't said anything on the matter, either, that I know of.

DSC08663

The mysterious "fifth tab"

With that, I don't want to get too far down the rabbit hole - this is Android Auto in its earliest publicly available incarnation. Six months from now, who knows what we might see (possibly almost nothing at all, but hey!). As it is, this is clearly a good jumping off point to get Google and some real Android functionality into your car, and it certainly makes some compelling arguments for its existence. I, for one, will watch with great curiosity as Auto evolves and the ecosystem around it [hopefully] grows.