When AT&T decided to overhaul its approach to research and development it called a few prominent venture capitalists in Silicon Valley and invited the start-ups they were investing in on a “speed date”.
Principal architect for technology and innovation Sanjay Macwan says each startup was given 15 minutes to present its idea to a panel of senior executives. If the panel thought the idea could assist in developing AT&T’s network infrastructure it would “matchmake” the start-up with one of its technology partners such as Ericsson or Alcatel-Lucent.
If the AT&T executives thought the start-up offers a useful consumer or enterprise service, the start-up would be referred to the Foundry – a hothouse environment where the start-up has up to 90 days to prove the concept. If successful, the product was put into in beta for 12 weeks before commercialisation was considered.
The programme has now been running three years and in that time AT&T has invested US $90 million in the Foundry (there are three sites – two in the US and one in Israel), has conducted over 500 “speed dates” and launched nine new services.
Macwan says most of the speed dating has taken place in the US, but a panel has convened for a pitching session from start-ups based in India via Telepresence. So does that mean Macwan would be open to a session with New Zealand start-ups via Telepresence? Yes, says Macwan, he would consider it.
AT&T is reaching out to start-ups because rather than be just a ‘dumb pipe’ it wants to cement its dominance further up the food chain. The strategy is to create a wide range of API (Application Programming Interfaces) upon which developers can build their applications.
The company’s objective this year is to have 10 billion API transactions per month. In 2011 there were 4.5 billion API transactions per month. In 2008 there were none. Currently developers pay a US$99 yearly access fee to use the API, however next year there will be a change to the pricing model.
In addition to encouraging start-ups, AT&T has launched the ‘innovation pipeline’ in order to tap its 260,000 employees for ideas. Any employee can submit an idea, which is then opened up for collaboration and/or election by other employees. Each financial quarter the ten ideas which have attracted the most votes are pitched to a panel of top executives.
Three of these ideas are then selected for either the Foundry programme or developed further at AT&T labs. Macwan says that in three years 44,000 employees have participated in the innovation pipeline, 133,000 employees have voted, and US$32 million has been committed in funding. The IP behind the employee’s idea is owned by AT&T.
Alongside these two programmes are the AT&T laboratories, in which engineers and computer scientists get paid a salary to come up with innovative ideas. When Computerworld visited the Shannon Laboratory in New Jersey, as part of an Asia Pacific media tour, we were shown APIs around voice recognition software that enabled speech to text and (almost) instantaneous language translation, and a security API that combined traditional passwords with biometric and location-based information.
Another group at the lab is developing devices that are akin to what is often referred to as “the internet of things”. A haptic steering wheel with 20 sensors is being designed to use the power of touch for car GPS systems for example. In addition to directing you to turn left via a screen display, the left-hand side of the steering wheel vibrates to let you know when to make the turn.
Other devices under construction are ‘Got My Stuff’, which uses RFID tags to ensure you keep track of personal items such as sunglasses, smartphones and your wallet, and ‘Shadowpuppets’ which projects a map from your phone onto a screen and enables you to investigate locations by waving your hand around in front of the image.
Does AT&T ever worry that someone else will come up with the same idea and put it into market before them? “No,” says Gerald Karam, a computer scientist. He points out that “three guys in a garage” don’t have the resources to enable the kind of capability that AT&T has. (Not to mention the marketing dollars that can be deployed to push new products into the market).
The API strategy goes hand in hand with AT&T’s push towards a more open network globally. It says it wants to ensure the APIs it develops can be used on any carrier’s network.
The company is also working towards closer partnership and collaboration with other telcos, recognising that multiple end points – even if you don’t own them all yourself – and interconnection is at the heart of telecommunications.
For example, Telepresence executive director Alan Benway says AT&T is working with BT and Orange to push Cisco’s videoconference solution. It has adopted a revenue share model, so that if an AT&T customer calls someone on the BT network, BT charges AT&T for the call – and vice versa. “We’re trying to recreate the PSTN for the video industry,” Benway says.
AT&T’s Global Network Operations Centre in New Jersey (pictured above) monitors 30 petabytes of traffic a day.
At its hub is a large room featuring 141 video wallboards that provide a high level view of the cable systems owned – or part-owned — by AT&T. The IP backbone includes 909,000 world-wide fibre-route miles, 214,000 wi-fi hotspots, 16 million broadband connections and more than 98 million wireless customers.
AT&T’s business outside the US is mainly in managed services for enterprise customers which have multiple global sites. AT&T has 38 datacentres located around the world – including a datacentres in Sydney. AT&T’s business in the Asia Pacific region grew by nine percent last year, but inAustralia and New Zealand it was just two percent.
• Putt travelled to New Jersey as a guest of AT&T.