Will Cue Make Apple’s Siri Creepier In a Good Way?

CueA really good and useful personal digital assistant has to understand you. Like the proverbial butler of victorian literature, your personal assistant needs to know you and your quirks and it has to have insight into your daily routines, your current appointment calendar, your upcoming planned trips, and so forth.

Lots of folks praise Google Now over Apple’s Siri because of Google Now’s deep reach into the places where you store your information. Google Now can scan the contents of your Google email addresses and pick out information related to items you’ve purchased and flights you’ve booked. It can track packages for you or alert you about flight delays before you’ve even thought to ask about such things. This type of access to your personal data sphere is a little creepy, but it makes Google Now a more useful and capable personal assistant.

A company called Cue, formerly known as Greplin, offered a service that mimicked the victorian butler of whom we spoke earlier. Cue’s application acted as a an accumulator of important information from all of your social media identities. The app could sift through your email, Facebook, Twitter, LinkedIN and other accounts to create a unified index and a central inbox. Having all this knowledge about you in one place gave it the ability to predict what you might be looking for, based on context.

Now Apple has acquired Cue and the speculation abounds about how Apple will use the Cue predictive reasoning abilities to amp up Siri. If Siri has the ability to scrape data from our social and email treasure troves, it’ll certainly make it (her?) a more capable companion. Of course, there’s always the concern that we’re giving up some privacy in exchange for this added functionality. But can you ever really have any secrets from your butler? Probably not. Former Cue users will have to wait and see how it all turns out, since the Cue service is no longer available.

Apple’s Siri and the Unfulfilled Potential of Task Completion

SiriIn a recent Techcrunch article, Dan Kaplan laments the unfulfilled potential of Apple’s personal assistant Siri. Kaplan makes the point that the vision of the original Siri creators was for the mobile personal assistant to become  a “task-completion engine.” Siri was about disrupting traditional search and replacing it with an artificially intelligent assistant that could understand and then actually execute tasks for us. Instead of just looking up answers on the Internet and feeding them back to us, Siri’s destiny was supposed to be that of an indispensable concierge in a box.

Earlier pre-Apple versions of the assistant, Kaplan points out, could order you a cab, make restaurant reservations, or even find out what bands were playing around town. Siri’s creators envisioned even more task-completion functions, such as the ability to rebook an airline ticket for you if your flight got canceled or redirecting a package delivery to your office if it arrived at your house while you were out.

Kaplan complains that Siri has stagnated. The mobile assistant remains mired in the technology of search. Google Now has surpassed Siri in terms of its predictive powers. Google Now, for example, shows you cards with information based on tidbits it gleams from your Gmail accounts and other sources. That means that Google Now can show you the status of an airline flight you’re on without you asking, or update you on the score of a recent game played by a sports team you follow.

Kaplan’s point about the potential power of mobile task-completion is a good one. Most virtual agent technology today is focused on understanding language and intent and providing a good response. Even super sophisticated cognitive computing systems, like IBM Watson (aka DeepQA) are designed to provide answers to questions. Task-oriented virtual agents might be just around the corner, and it will be interesting to observe their impact on the marketplace. What will it mean when we have virtual agents that are not only able to converse with customers, but that can actually process customer transactions? We’re likely to find out one of these days.