Imagine this: you’re in a meeting, and your phone buzzes to indicate that you have an important call. You excuse yourself to take the call and a synthesized voice lets you know that a judge just handed down a surprise verdict in a patent infringement suit between two of your competitors. You’d accounted for several potential scenarios, but not the one that unfolded.
Your digital assistant reads every email you send and receive every document you read and write. It knows that you sent your assessment of the company’s vulnerability to your CEO. It determined there was a high probability that you would need to take action immediately and interrupted an important meeting to inform you the moment the news hit the wires.
This scenario may still sound like science fiction, but it’s getting closer, and the stakes are getting higher. The increasing amount of data from outside the company—from customers, reviewers, business partners and competitors—confronting over-extended executives makes deriving meaning from the onslaught more than a visionary goal.
Data analytics is evolving from advantage to necessity, said Erick Brethenoux, IBM’s Director of Business Analytics. With data analytics, this information is now “loud, readable and understandable,” he said. “If you’re not taking advantage of, or even taking into consideration, that voice on the outside, then you’re going to be out of the picture very quickly.”
The here-and-now analog for this vision is Siri, Apple’s voice-controlled digital assistant for iOS devices. Siri makes it easy to set reminders, search the Web and dial the phone by voice. It also makes it easy to imagine a future when digital assistants with considerably more mental prowess serve as executive assistants to help business people handle information overload.
It’ll take a lot of work to get to an enterprise Siri. Enterprises must be able to glean information from vast amounts of data, integrate the information into business processes, and then make it useful in workers’ daily lives.
The ideal enterprise digital assistant combines a Siri-like speech interface, a Google-like knowledge of your interests and an IBM Watson-like ability to reason. This combination will allow it to act much like a human assistant.
The Need to Build Context in a World of Data Silos
But before we can use enterprise digital assistants to help us sift through all this information, the industry must solve the silo problem. A lot of business data is buried in an array of systems like Salesforce, SAP, Box.net and Workday, said Michael Driscoll, CEO of Metamarkets, Inc., a company that provides data analytics for mobile and online advertising. The first challenge is to create an interface that interoperates over all these services, he said.
Many vendors in the enterprise search space are working toward unified information access. Unifying search across an enterprise’s varied data stores is useful in its own right. It’s also a prerequisite for deriving insights by correlating information.
You can’t really understand data without knowing its context, said Judith Hurwitz, president and CEO of consulting and market research firm Hurwitz and Associates. You get a much deeper and broader picture if you get beyond silos and correlate disparate pieces of information, she said. ”As you get a handle on this, it starts to transform how you look at the world and how you use data to make decisions,” she said. “Some of those decisions may actually be in real time.”
Key to making use of correlated data — and an important component of an enterprise digital assistant—is the interface between the user and the data. Siri’s power comes from being an effective (for the most part) intermediary between the user and the apps that deliver the data and services.
If the increasing use of mobile devices and Web apps represents the consumerization of the enterprise, designed interfaces will represent the Apple-ification of the enterprise, said Driscoll. We need designed interfaces that make it easy to interact with and get at the information we need to do our jobs, he said. “Enterprise software is still too hard to use – interfaces for business users [accessing] enterprise data are woefully underdeveloped,” he said.
From Visualizations to Tailored Recommendations
While well-designed interfaces allow users to access information, a lot of back-end processing is also required to correlate and package the information in the first place. Data visualization makes dense and complex information readily accessible.
Pharmaceutical researchers have been using it for years to search for and design valuable compounds, said Hurwitz. There are too many variables for researchers to keep track of without visualization. As data analytics spreads across industries, the same will be true in many more areas, said Hurwitz. “Data visualization, especially when you’re dealing with huge volumes of data, is going to be crucial,” she said.
But sometimes data visualization doesn’t go far enough. An ideal enterprise digital assistant would look at the data for you, summarize it and make recommendations for action. It’s like having turn-by-turn directions versus looking at a map, said Driscoll. “How do we create a layer of intelligence over all this data so we’re not just presenting people information [but] intelligently suggesting what they do next,” he said.
Accomplishing such a lofty goal requires using data analytics tools on the user who is accessing the data. It will mean understanding her job function, her business network, her routine, the goals and histories of the projects she’s working on, her habits and preferences, her location, her schedule and those of her colleagues.
In short, it means building a model of her information needs that can filter, package and time the delivery of information, said IBM’s Brethenoux. This will allow a digital assistant to “only give [information] to me if I need it, and then when you do, be ready to give me more if I ask you some questions if I need to explore that a little deeper to understand it,” he said. “And then give me the scenarios of what I should be doing and let me make the decision.”
Such a vision verges on artificial intelligence. IBM’s Watson is a big step in this direction. Ultimately, we’ll have back-end artificial intelligence that will know you so well that the balance of power and resources between back end and front end, which is about 50-50 today, will be more like 90-10, said Brethenoux. This is because the back end will process information so effectively that there’s less for the interface to do, he said.
The Way Forward: Personalized and Predictive
Such an enterprise digital assistant will take a lot of effort to develop, but there will be useful milestones along the way. The key is not trying to tackle it all at once, said Driscoll. “My advice to entrepreneurs is to start small,” he said. “Siri for the enterprise is a ‘boil the ocean’ problem.”
One approach is interfacing with specific tools like Salesforce, said Driscoll. “Start with the data people care most about, sales and revenue data, and create a basic interface for a mobile phone,” he said. “Price data is what CEOs care a lot about: who sold what, how many, at what price?”
Another approach is for resource-rich companies to go all in, à la IBM with Watson. Ray Kurzweil, the famed artificial intelligence researcher recently hired by Google, plans to turn Google’s ubiquitous search into a powerful digital assistant. His vision is of a highly intelligent Google search that tracks what people read, what emails they send and what conversations they have, and develops detailed profiles of their needs and interests — think Google Now on steroids. “It’ll be constantly surveying all of the new knowledge that comes out every minute and bringing things to your attention that it thinks you will want,” Kurzweil said in a videotaped interview with Singularity Hub in January.
Enterprise digital assistants may be closer than we think, said Brethenoux. The adoption curve is becoming elongated, with bleeding-edge users deploying technologies fresh out of the laboratories, he said. “It’s not 20 years out. I’m sure of that,” said Brethenoux.
“I envision some years from now the majority of search queries will be answered without actually you asking,” said Kurzweil. “It’ll just know that this is something you’re going to want to see.”
Eric Smalley is a freelance writer in Boston. He is a regular contributor to Wired.com. Follow him on Twitter at @ericsmalley.