To create an app that allows deaf people to use smartphones, entrepreneur Kunal Batra needed software that could turn speech into text. There was no way his startup, General Machines, would want to develop such complicated software from scratch.
So Batra tested Google Voice and Twilio but eventually decided that AT&T’s speech-to-text program Watson—refined and improved over decades at the telecommunication company’s research labs—performed the best. His phone app, Deaftel, now operates by accessing the software on AT&T servers free of charge, although Barta expects he’ll eventually pay about a penny a minute.
Batra’s decision counts as a small victory for AT&T, which is aggressively seeking outside help in managing what it calls “one of the largest, fastest” waves of technological change ever: the rise of the mobile Internet.
Even though it has a $1.2 billion research budget, an R&D staff of more than 1,300, and a legacy of invention dating back to Bell Laboratories, those resources now seem small compared with the army of coders who have begun building mobile apps for iPhones and Android devices operating over AT&T’s network. To keep up, AT&T is pursuing a new strategy designed to capture developers’ attention. It is holding programming contests, spent $100 million to open several “innovation centers” where its staff can work with startups, and, most significantly, has begun distributing the fundamental currency of the app economy—cheap and easy access to useful data, including streams of information about what its customers are doing, as well as software like Watson.
For AT&T, that is a step toward outsourcing product innovation, an activity it has historically undertaken itself, either with in-house engineers or by putting out contracts with design requirements. Now, at least with mobile apps, AT&T’s money and efforts are also being steered into the technology adaptations, image polishing, and revenue sharing required to maintain a healthy “ecosystem” of outside collaborators who it hopes will innovate for it.
Other established companies, including FedEx and Ford, are also cracking open data once considered strictly proprietary. FedEx has opened access to package-tracking data (one developer created an app to gauge carbon dioxide emissions for every parcel sent). Ford has given some developers access to data from its cars’ central computers, hoping they’ll come up with clever ideas, like apps that track the fuel efficiency of a person’s driving (see “Ford Bets on the Digital Car”).
For AT&T, an irony is that it helped launch the era of apps when Apple introduced the iPhone in 2007. AT&T was the exclusive carrier, but it was Apple that vaulted to become the most highly valued company in the world. Developers stormed onto Apple’s platform, writing more than 600,000 apps for the phone.
That shift is in some ways dangerous for AT&T’s business. Take Skype, the software owned by Microsoft for making calls over the Internet. More and more people are using Skype’s mobile app to make calls from their phones, but because it is powered by streams of data, not a voice plan, AT&T may not make as much money on those calls. Worse, such apps dilute the phone giant’s brand. People now say “I’ll Skype you” but never “How about an AT&T call?”
“They are slowly being pushed out and don’t want to become a simple pipe,” says Fima Katz, CEO of Tiggzi, a mobile-app development company that has worked with AT&T. “Everyone [is] eating their lunch.”
To adjust, the company has been racing to throw open its systems to developers and fast-track apps that use AT&T data. The technology involved is known as application programming interfaces, or APIs. These are the programs that let AT&T give outside software applications real-time access to the data, or services like Watson, it wants to make available. AT&T now offers 79 APIs that process five billion “calls” each month (each call is a request from an outside program for data), up from 300 million calls per month in 2009, according to Jon Summers, AT&T’s senior vice president of growth platforms.
The API that permits apps to connect to Watson, released this summer, is AT&T’s most recent addition. Others share a phone’s geographic location, permit payments, or can encrypt a document. Yet another works with AT&T’s U-Verse television service; developers at Miso, a startup, used it to create an app that lets people interact with friends who are watching the same show.
Summers says the APIs not only are helpful for stoking interest among outside software developers but also will speed projects inside AT&T. He estimates that using them will help AT&T with its aim of accelerating its average product development timeline to six months, from 18 or 24 months.
Compared with many companies, AT&T’s efforts are still incipient. The website Klout, which gives people an estimate of their “influence” on social networks like Facebook, said earlier this year its API was getting over a billion calls each day. Twitter has gotten more than 15 billion in one day.
If AT&T is successful, it may find that fueling the inventiveness of other companies has drawbacks. Twitter, for instance, recently restricted who can use its API after deciding there was a bit too much innovation going on—too many companies were using data about tweets to build products that didn’t benefit Twitter (see “Startups Worry that Twitter and Facebook Are Blocking Their Way”).
Some of AT&T’s APIs could similarly allow developers to compete with the company. State governments and the Federal Communications Commission, for instance, are required by law to provide communications services to deaf people. Phone companies including AT&T currently employ human operators to relay deaf people’s texts, billing governments at rates exceeding $1 a minute.
Batra, the Deaftel entrepreneur, says his pitch is exactly that: his AT&T-powered app will do it for much less.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
How to fix the internet
If we want online discourse to improve, we need to move beyond the big platforms.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.