Smart apps that learn and adapt to their users spur brave new tech world
Technologies that end up improving lives and changing culture often seem trivial when they're brand new.
Blogging, for example, began as a way to keep an online diary but has evolved to a medium that's transforming journalism and business.
Social media started out as a way for teenagers and college students to flirt with one another, but it has become one of the primary ways people discover content online.
Likewise, a brand-new generation of smart apps may appear to be limited toys for geeks and productivity enthusiasts. In fact, they represent first steps toward the future of all human-machine interaction -- a future in which we hold conversations with our computers and they get to know us, learn how to suggest things, solve some of our everyday problems and go out into the world doing chores on our behalf.
This new category appears to contain far-flung and divergent capabilities. Generally, they have the ability to do one or more of the following:
- Learn user context and preferences. By paying attention to our choices and behaviors, software learns to predict what we'll want.
- Rely on artificial intelligence to make decisions. Algorithms enable software to discern between relevant and irrelevant incoming information.
- Contact users to provide contextually relevant information. Rather than waiting for us to search for something, they can buzz our phones with answers to questions we haven't asked.
- Act proactively. Instead of waiting for us to take action, they take action for us.
- Automate tasks. Either users or the software can set up if-then commands like programmers do, and they can do so across different applications and services.
- Communicate as the user. Software learns to know who you'll communicate with and what you'll say, then does it for you. To the recipient, it appears as though the message comes from you. When both parties are using software agents to communicate, it's just software talking to software.
- Facilitate users' actions. Agents figure out what you'll want to do and get it ready for you. By pressing a single button, you can tell the system to do something that would otherwise be a multi-step process.
- Act with agency on the user's behalf. Software does something for you without asking permission or informing you in advance.
It all sounds science-fictionish and Star Trekky. But these capabilities are already available in free apps and Web-based services and will increasingly be added to most of our apps, services, websites and consumer products.
Some of these services act as a user interface for existing apps and services, and others function as a kind of glue.
Apple's Siri, for example, is mostly a user interface for a variety of services. You talk, and something is added to your calendar. You talk again, and you set a reminder in a different app. At Mobile World Congress last month, a company called Artificial Solutions introduced a Siri-like app called Indigo for non- Apple devices -- specifically Android and Windows Phone devices.
It's not immediately clear that everybody will embrace "smart" apps, services and features. I suspect that smart features will start showing up in so many places all at once that users will feel overwhelmed and confused, unable to understand which services are doing what.
That's why I'm a big fan of an iOS app called EasilyDo, which does something I like to call "facilitated reality." Instead of just doing things for you, it notifies you about things you probably want to do, and then does them for you only after you tap the "Do It" button.
It's like a caddy that discovers the best nearby golf course, drives you there, carries your clubs and tees up the ball for you. All you have to do is hit the ball (or decide not to).
As an example, EasilyDo reads your friends' Facebook posts and can tell the difference between good, bad and indifferent news. Let's say a friend announces that he was promoted at work. EasilyDo recognizes the nature of the post, notifies you and tees up a "Congratulations!" comment to the friend's post. You just have to press one button for "your" comment to be posted by "you." (You can edit the message if you want to.)
The app can do all kinds of powerful things. It can tell you when to leave your office in order to be on time for your next meeting, or remind you to pay bills based on incoming email.
This week, EasilyDo rolled out a new browser-based component in beta called EasilyDo Builder, which lets you refine and customize these "facilitated reality" actions.
You do the refining on the Web, but the iPhone app performs the actions.
The new EasilyDo Builder reminds me of IFTTT, since it lets you glue together apps and services with commands to take action.
EasilyDo is more limited in scope than IFTTT, but nontechnical users will find it a lot easier to use and manage.
The founders of EasilyDo told me recently that they intend to add "agency" to the app; it will be able to order flowers and make restaurant reservations to save your bacon on Valentine's Day -- that kind of thing. (The app already has a credit card field so it can pay for things on your behalf.)
EasilyDo recently announced a commitment to add a new capability every week. (The most recent is that you can use search criteria to selectively add incoming emails to Evernote.)
Another app, called Grokr, uses a scrolling interface similar to EasilyDo's. The two apps are often compared to each other, but Grokr is very different from EasilyDo. Grokr focuses on traffic and weather, it shows you news and other articles based on your interests, and it lists trending stories. I have found Grokr not particularly useful. The information seems generic, and it doesn't seem to learn.
And then there's Google
Google produces the most mind-blowing intelligent agent services for consumers from a technology perspective. Unlike EasilyDo's services, Google's tend to be informational rather than action-oriented.
Of course Google Now is probably the single most impressive piece of agent technology generally available. If you have the right kind of Android device, Google Now can replace search, giving you answers rather than search results. Best of all, it understands everyday language and learns from your results. Over time, it suggests things based on your preferences.
It's reasonable to assume that Google Now will show up everywhere, including on the Google Search page, in the Google Chrome browser and even on the iOS platform as an app.
Google this week released its Field Trip app on the iOS platform. (It had previously been available only on Android since September.)
Field Trip is interesting for its location-awareness and proactivity. As you're walking around, it can buzz your phone to tell you about interesting things nearby. It learns your location from your phone and then finds information about that location from Zagat, Scoutmob, Cool Hunting, Yesterland, Curbed and Thrillist.
An app called Spindle is vaguely similar to Field Trip, but it only works in six U.S. cities: Austin, Boston, Chicago, New York, San Francisco and Seattle.
Another interesting app in this category, called Tempo, replaces your iOS calendar with a smarter one.
Tempo can snatch information from your iPhone's Mail, Contacts and location to improve what happens with your Calendar. For example, if you have a call, it will grab the number for you. If you have a meeting, it can give you details about the person you're meeting with (from Contacts and Linkedin).
Tempo is also designed to give you estimates of drive times, even allotting time for parking.
Tempo was created by some of the same software people who created Siri, and it really does things Siri should do by now.
The new intelligent agent apps and services are fun to play with, can keep you informed, and can make you far more productive. All of them are free and easy to use, but they're not trivial.
In fact, these products represent the future of computing.
Over the next few years, almost every app we use and every website we visit may function less like a machine we're using and more like a person helping us to do our work and live our lives.
If we rely more and more heavily on algorithms to make decisions, communicate with our friends and otherwise act on our behalf, will we lose the ability to do things for ourselves?
In other words, as our machines get smarter, will we get dumber?
I wish there was an app that could tell us the answer to that question.
Mike Elgan writes about technology and tech culture. You can contact Mike and learn more about him on Google+. You can also see more articles by Mike Elgan on Computerworld.com.
Read more about mobile apps in Computerworld's Mobile Apps Topic Center.