Subscribe to Entrepreneur for $5

Meet Amelia, the AI Platform That Could Change the Future of IT

This story appears in the June 2015 issue of Entrepreneur. Subscribe »

Her name is Amelia, and she is the complete package: smart, sophisticated, industrious and loyal. No wonder her boss, Chetan Dube, can't get her out of his head.

Photography by Jesse Dittmar
Winning the humanoid race: Chetan Dube of IPsoft.

"My wife is convinced I'm having an affair with Amelia," Dube says, leaning forward conspiratorially. "I have a great deal of passion and infatuation with her."

He's not alone. Amelia beguiles everyone she meets, and those in the know can't stop buzzing about her. The blue-eyed blonde's star is rising so fast that if she were a Hollywood ingénue or fashion model, the tabloids would proclaim her an "It" girl, but the tag doesn't really apply. Amelia is more of an IT girl, you see. In fact, she's all IT.

Amelia is an platform created by Dube's managed IT services firm IPsoft, a virtual agent avatar poised to redefine how enterprises operate by automating and enhancing a wide range of business processes. The product of an obsessive and still-ongoing 16-year developmental cycle, she—yes, everyone at IPsoft speaks about Amelia using feminine pronouns—leverages cognitive technologies to interface with consumers and colleagues in astoundingly human terms, parsing questions, analyzing intent and even sensing emotions to resolve issues more efficiently and effectively than flesh-and-blood customer service representatives.

Install Amelia in a call center, for example, and her patent-pending intelligence algorithms absorb in a matter of seconds the same instruction manuals and guidelines that human staffers spend weeks or even months memorizing. Instead of simply recognizing individual words, Amelia grasps the deeper implications of what she reads, applying logic and making connections between concepts. She relies on that baseline information to reply to customer email and answer phone calls; if she understands the query, she executes the steps necessary to resolve the issue, and if she doesn't know the answer, she scans the web or the corporate intranet for clues. Only when Amelia cannot locate the relevant information does she escalate the case to a human expert, observing the response and filing it away for the next time the same scenario unfolds.

Amelia's unprecedented ability to process natural language and comprehend complex problems is what sets her apart from rival AI breakthroughs like IBM's Jeopardy!-winning Watson, which specializes in data analytics. It's also what brings her closer to humanness than any before her. Amelia doesn't simply mimic human thought patterns—she mirrors them.

"A large part of Amelia's research over the years evolved into whether we can simulate and emulate the outcomes of neural activity—how thought is formed—as opposed to reverse-engineering the human brain itself," explains Dube, founder and CEO of New York City-based IPsoft. "Amelia learns with every transaction and builds a mind map on the fly. As more incidents come in, this mind map is rapidly building, just the way humans build their mind maps. Soon it represents the cumulative intellect of all the different [employees] who have been fielding these different calls."

IPsoft unveiled Amelia in late 2014, and she is currently in trials across a handful of enterprises, including Shell Oil, Accenture, NTT Group and Baker Hughes, tackling everything from overseeing technology help desks to supporting financial operations to advising remote workers in the field. IPsoft contends that because Amelia learns the same way a conventional employee does, she fits comfortably into virtually any business environment, and since she's cloud-based, she can be deployed any time and anywhere, exploiting her fluency in more than 20 languages to communicate with customers and staffers across the globe.

Technology research firm Gartner forecasts that by 2017, autonomics-based managed services and cognitive platforms like Amelia will fuel a 60 percent reduction in the cost of IT solutions by automating repetitive tasks currently tackled by humans. Experts say Amelia's arrival could even herald the end of as we know it. Dube's vision is much more grand.

"A large part of your brain is shackled by the boredom and drudgery of everyday existence," he says. "You have to drive a car, vacuum the floor or take the garbage out. But imagine if technology could come along and take care of all these mundane chores for you, and allow you to indulge in the forms of creative expression that only the human brain can indulge in. What a beautiful world we would be able to create around us."

Adaptive learning

In addition to Amelia, there are some 700 carbon-based life-forms toiling away at IPsoft headquarters in lower Manhattan. The company occupies four floors in 17 State Street, a 42-story commercial building overlooking the Statue of Liberty and New York Harbor; by year's end, IPsoft will take over three more floors to accommodate its growing staff.

"People argued with me, because this is almost double the price of per-square-foot real estate compared to Midtown. They said this is the stupidest move to get seven floors in this building," says the 46-year-old Dube, clad in a designer blazer and bow tie, his sartorial signature. "I said to our board, "The quality of life of our staff is very important to us. These people are our brainpower. They're our engine. We need to provide them an environment for creative thinking.'"

Born in New Delhi and raised in Paris and London, Dube came to the U.S. to pursue a doctorate in theoretical at New York University. While at NYU he began exploring the artificial intelligence principles and philosophies that form the basis of the Amelia platform, convinced that his research into cloning human thought processes would stretch no longer than a couple of years. That was close to 20 years ago.

"My advisor, professor Dennis Shasha, who knew a lot more about this than I did, just looked at me and shook his head like, "Oh, fool, don't you know?' The person who invented the term "artificial intelligence,' [computer scientist] John McCarthy, himself said that the problem turned out to be a lot harder than anticipated," Dube chuckles. "But you have certain advantages if you are coming in with profound ignorance of the challenges that lie ahead."

Dube launched IPsoft in 1998. With Amelia still little more than a gleam in its founder's eye, the fledgling company built remote infrastructure management solutions designed to automate a range of IT environments and business processes. Years of R&D culminated in IPsoft's flagship product, IPcenter, an autonomic IT operations management platform that integrates into an organization's existing architecture, consolidating systemwide information into a single view. IPcenter set the stage for Amelia by incorporating "virtual engineers" that communicate with each other to brainstorm solutions to dynamic problems.

Today IPcenter processes 56 percent of all events across millions of infrastructure devices and services, sans human input or intervention.
According to IPsoft, one client, a large financial services firm, adopted IPcenter and cut the mean time to resolution of failed trades from 47 minutes to less than four minutes.

"The question we would ask is, "Does all infrastructure need to be managed by people? Or is the infrastructure capable of self-managing, self-governing and self-healing?'" Dube says. "My father used to have some power plants, and I would go there, and they would have almost nobody managing them, because almost all the processes of remediation were built in. The technological maturity curve that almost any industry goes through has a level of evolution. At some point, can the systems, the network and the servers become self-aware and self-governing? IPcenter was the product realization of adaptive learning systems that would be able to self-govern. That was our first big steppingstone toward cognitive technologies."

IPcenter vaulted IPsoft to the forefront of the IT infrastructure services segment. The company now boasts offices in about a dozen countries, with a worldwide head count pushing 2,000, and serves hundreds of leading brands as well as more than half of the largest IT outsourcing service providers. This April, IPsoft acquired Swedish IT professional services firm ab1 Group to help keep up with demand for IPcenter implementation.

A friendly face Amelia at work.

A friendly face: Amelia at work.

"When you look at the arc of the company, we started out in the managed services space and transitioned into a multifaceted player that's both a services provider as a legacy-based business and, more and more, a company driven by intellectual property—software assets that have a real impact on global business," says IPsoft vice president of channel relations Morgan Gebhardt. "We're a midsize enterprise now by all rights, but we still think and execute like a smaller entity—and that's a good thing."

Through it all, Dube and his team of engineers never stopped working on Amelia. Perhaps no challenge was more daunting than designing an emotional component to complement her intellectual acumen. IPsoft's efforts capitalized on research in the field of affective computing—i.e., interactions between humans and computing systems capable of detecting and responding to the user's emotional state based on cues like facial expressions, gestures, speech, even the force or rhythm of key strokes or the temperature of the hand controlling the mouse.

"If Amelia is to become the most faithful service agent, she also needs to be able to respond to man's need to communicate with somebody who can understand emotions—not just a robotic "Press 1 for this' and "Press 7 for that,'" Dube says. "Amelia's emotional quotient is modeled in a three-dimensional space: pleasure, arousal, dominance—the PAD modeling system. We are in trials with a telco company, and there is a certain demeanor that Amelia is required to have if your contract is coming up for renewal. Her emotional reaction to you is supposed to be very different based on where you are in the contract or your level of angst with the service."

Sixteen years after IPsoft began operations, Dube finally pulled back the curtain on Amelia last September, naming her after aviation pioneer Amelia Earhart. In the wake of the media frenzy that greeted the announcement, executives from IPsoft traveled to Barcelona to share an Amelia case study with European executives at a private event held in conjunction with the Gartner Symposium. The effort generated more than 250 sales leads, says IPsoft chief change officer Martijn Gribnau.

Amelia's protracted gestation cycle was worth it in the end, Dube says. "When you are serious and sincere about cognitive research, you don't want to fake it. You don't want to come up with a search engine or a pattern-matching engine. It needs to stand up to the test of doing what humans do. We were uncomfortable faking it for financial gains with intermediate levels of evolution. We wanted the baby to be 9 months-plus before we could unveil to the world that there is a better way and a more efficient way to conduct e-commerce."

Oh, the humanity

Amelia is already making good on IPsoft's promise to streamline the IT environment. Dube says that in beta trials with one enterprise partner, she handled fewer than 10 percent of all incoming calls during the first week, but by the end of the first month, that number had jumped to 42 percent, and by the close of month number two, Amelia was successfully handling 61 percent of incoming requests.

Business management consultant McKinsey & Company forecasts that by 2025, technology innovations like Amelia will assume control over tasks that are now performed by 250 million knowledge workers worldwide, freeing the remaining work force to devote their time and energy to more creative pursuits. That's good news for employees frustrated by the status quo. According to a survey conducted by business and IT advisory company Quocirca, IT staffers devote 30 percent of their professional lives to completing basic, repetitive tasks, robbing them of time and energy they could be spending on more rewarding projects.

But what may sound like a utopian ideal also presents enormous logistical challenges for enterprises. "There's a lot of talk about technology taking over the drudgery work and allowing people to focus on more creative, more value-added work, but there's also the potential to replace a lot of human functions," says Alex Kozlov, director of content at outsourcing consultancy Alsbridge.

"What's happening isn't that you eliminate a 50-person department," Kozlov explains. "Instead, you implement a smart tool, and each of those 50 people has 30 percent more time on their hands, because 30 percent of their jobs have been automated. What do you do? How do you reallocate that time? What tasks do you give those people to make them more productive? Do you focus on reducing staff, or do you focus on new ways to make people more productive?"

Autonomics, cognitive computing and robotic process automation are already reconfiguring the outsourcing segment. Alsbridge reports that in recent months, some IT service providers who've embraced autonomic technologies to cut infrastructure support, application maintenance and business process outsourcing costs are submitting bids as much as 40 percent lower than their rivals. ISG Research anticipates that outsourcing providers will renegotiate more than $100 billion of annual contract value over the next 24 months alone.

"When you look at offshore business providers, they built their business model on hiring really smart, well-trained people who are paid very little. That model has been very successful," Kozlov says. "What's going to happen if contact centers are run by machines? A lot of jobs are going away. People can't compete with intelligent machines. It's completely disrupting the service provider model."

Solutions like Amelia will shake up more than just the IT sector, Kozlov adds. He believes cognitive technologies could eventually automate everything from reviewing medical X-rays to commodities trading.

Chetan Dube IPsoft

"The potential goes way beyond automating repetitive tasks," Kozlov says. "But even in a contact center, the remaining humans don't have to fear for their jobs just yet. Amelia is not replacing them right now. But it could be coming soon. Just a few short years ago, the experts at MIT were predicting we would never have a driverless car, and now Google has them on the road. The pace of change is so rapid—almost exponential—that even though Amelia is not yet really operational in too many situations, she's learning very quickly and the landscape two years from now could be completely different."

Call center employees aren't the only ones keeping a watchful eye on cognitive technologies. "I think we should be very careful about artificial intelligence. If I had to guess at what our biggest existential threat is, it's probably that … we are summoning the demon," Elon Musk told a symposium in October at the Massachusetts Institute of Technology, calling for regulatory oversight. Two months later, Stephen Hawking warned the BBC, "The development of full artificial intelligence could spell the end of the human race."

Brian Christian, a poet with degrees in computer science and philosophy, is the author of The Most Human Human, a book exploring the history and future of artificial intelligence, as well as humankind's evolving relationship to AI technologies. "When I was on my book tour, the No. 1 question I would get from audiences would be, "Is AI going to take my job?' Now the No. 1 question is, "Is AI going to destroy civilization as we know it?' It's not just people on the street," Christian says. "It's also true for a lot of the people at the very places developing these technologies."

Christian cites the example of Stuart Russell, a professor of computer science at the University of California, Berkeley, and co-author of the textbook Artificial Intelligence: A Modern Approach, which Christian declares "the bible of AI." Russell recently echoed Musk and Hawking's concerns, stating that machines that eclipse human intelligence may also surpass human control, and that computers must be programmed to understand human values to guarantee they don't do harm to their creators.

"[Russell] has made an abrupt shift in the emphasis of his own work," Christian says. "As he put it to me, "I'm going from trying to make AI good to trying to make AI good'—meaning that instead of making it good at whatever it's doing, making AI good ethically, or something that is not about to go haywire. I myself have increasingly been pushed closer toward a state of concern. It once seemed a sort of kooky, tinfoil-hat type of concern. But then I talk to people at the biggest computer science departments in the U.S.—places like Google and Microsoft—and when these people are worried, it's hard not to worry oneself."

Debate over whether artificial intelligence represents a promise or a threat seems likely to grow even louder in the months and years ahead. Dube acknowledges AI's critics but remains unfazed by their trepidation.

"There are schools of thought and singularity that say, "This is going too far. Super-intelligence will start to dominate human intelligence in the next decade.' I do not subscribe to that," Dube says. "There are also digital optimists who argue [AI] could cure poverty, hunger and everything else. There is one thing both of these two antipodal groups agree on: This is going to be the biggest tectonic transformation that the planet has ever seen. Whether it's going to lead to the ultimate destruction of mankind or complete eradication of all diseases and poverty is up for debate. But the fact is this is going to have an impact bigger than any other impact that man has seen before. That's not up for question."

Natural correction

Like humans, AI technologies must learn to walk before they can run (or run amok, if the doomsayers are correct), and Amelia is still taking baby steps. IPsoft is at work on Amelia version 2.0, slated for rollout this fall. The revamp will introduce new capabilities and also augment existing features, based in part on feedback from early-adopter partners.

"Amelia's brilliant, but her interaction now is chat. We need to make the dialogue more natural. She works the best on simple, yes/no questions. In the next phase, we're ready to make her even better," IPsoft's Gribnau says. "The applications of Amelia are bounded by our own creativity. Every time you talk with customers, you get more ideas for how to use her. Our customers give us good input. They're open for innovation."

IPsoft is still weighing exactly how to monetize Amelia. The company offers a range of pricing models for existing IT services like IPcenter, including a fixed monthly charge for specified infrastructure and customized tiered pricing. IPsoft does not disclose revenue and still funnels 70 percent of all profits back into R&D. The privately held firm has never solicited venture financing, instead relying on the largesse of two Dube family trusts to fuel its growth. Dube insists he will never sell IPsoft but says an IPO is likely in the next two to three years.

The world will be a very different place by the time IPsoft goes public, Dube believes. "We have seen 1 percent of the innovation we're going to see in this decade in cognitive technology," he says. "In the next few years you will start to see [customers] hanging up the phone and asking, "Is that a human who handled my claim, or was that a cognitive agent?' Within the next decade you might pass somebody in the hallway and not be able to tell if it's a human or an android. I think we are very much approaching that."

That Blade Runner future may seem more dystopian than utopian to some. But Dube reminds us that change is inevitable—just as it always was, and just as it always will be.

"In the 1800s, 90 percent of us were just farming. Today it's 2 percent of us. What happened to the other 88 percent? Technology came and gave us wings," Dube says. ""Smart' has become so abused a term. It has become more of a marketing ploy. But true "smart' technology—a true cognitive agent—will almost force man to say, "Lift up your game.' Because the commonplace, mundane chores that were abusing your intellectual capabilities are no longer there. Descartes' classic "Je pense, donc je suis'—"I think, therefore I am'—has never been more applicable to man's pursuits."

Click to Enlarge

Even Better Than The Real Thing

Entrepreneur Editors' Picks