App Intelligence

It's a deal. Have your wireless agent call my wireless agent in the morning.

In the future, you won't need “your people” to contact “their people” to schedule meetings, purchase new equipment or even close multimillion-dollar deals. Your wireless device simply will contact another wireless device to do these things in less than five minutes via a wireless agent app using little bandwidth.

Think this is ridiculous? Wireless agents and other intelligent apps are being tested in labs today and just may be coming to a device near you sooner than you think.

In 10 years, we may be walking around with wireless devices embedded with app intelligence that knows you, what you like for breakfast, the highways you take to work and your kids' daily schedules. Devices will know who you are, even what you want, and act accordingly. App-intelligence technology will enable smart devices to perform tasks and transactions on your behalf. This could include sending an SMS telling your spouse that you're stuck in traffic.

Your personal assistant should be afraid. Very afraid.

Astute Agents

Agent software, today only available for the fixed world, makes decisions and takes action on a user's behalf, with or without his being present. Wireless intelligent agents not only will act on a consumer's behalf but also will have the ability to act independently, learn, adapt or even evolve — all without having to maintain a constant wireless connection.

Apps for network-based wireless agents include data collection; searching and filtering; monitoring information; targeted information dissemination; and m-commerce transactions.

For example, an agent could look up the Kansas City Chiefs' schedule and ticket prices, find the best seat for a game, make reservations and pay with a credit card.

M-commerce also could take place between agents. For example, if a user wants to buy a car, he could tell his agent to create preferences and provide a price range. At the dedicated host, the agent would haggle with agents selling a car. If a potential match were found, the agent would report to the user and even could complete the sale.

Consumers could program an agent to wait as long as it takes for certain information to become available. Also, devices don't have to stay connected to the network until an agent returns with a report.

Personal agents also can interact with each other. If, for example, a user wants to schedule a meeting with colleagues, he could send his agent to interact with the agents of those he wants to attend. The agents could exchange information, negotiate and establish a meeting time.

“Impulse,” a project created by Massachusetts Institute of Technology Media Lab (www.media.mit.edu) researchers Joan Morris and Jim Youll employs wireless agents to conduct price comparisons and negotiations for consumers. M-commerce meets brick-and-mortar in a system of buyers' and sellers' agents representing individual consumers and retailers negotiating via wireless devices.

A wireless device can negotiate for a product with an online or brick-and-mortar store or add the product to a wish list for ongoing negotiations. At any time, consumers can add items to the list, noting preferences such as warranty terms, merchant reputation, availability, time limit for the purchase and preferred price.

As a consumer visits stores and malls, the intelligent agent will engage nearby merchants in a silent exchange seeking the items on the list and opening sale negotiations, alerting the user if a deal is reached. The merchants' agents negotiate sales terms, considering loyalty, probability of an immediate sale, and the age and shelf life of goods in stock, according to the retailer's guidelines.

Preference-Filter Future

Today's Internet filters require you to fill out fields such as name, address and phone number so a company can best meet your needs. Tomorrow's wireless preference-filter apps will definitely know your name and may be able to learn your needs intuitively.

Amazon.com has been using a preference filter to suggest books and music to consumers based on their past purchases. In the wireless world, however, preference won't arrive for a while, says Jeff Belk, Qualcomm (www.qualcomm.com) senior vice president of marketing.

“Outside of Amazon, there's not a big list of preference-pushed things that are actually monitoring your behavior and giving recommendations today,” he says. “There's a great opportunity for the wireless carrier to build value with customers and a great opportunity for customers to get new capabilities.”

With preference filters, for example, a phone would know whether the user is a fan of the Kansas City Chiefs or Oakland Raiders. The device could find a streaming audio or video feed of the Oakland game — even the one with the user's favorite announcers.

Wireless preference filters of the future may take one of four forms. With profile filtering, you describe your interests by choosing from a list or entering keywords, and the software rejects anything that doesn't match. Collaborative filtering predicts your preferences by comparing your likes and dislikes to those of other people. Psychographic filtering predicts your likes and dislikes based on a psychographic profile derived from a questionnaire. Adaptive filtering learns as it goes by asking you to rate things or by monitoring your actions.

Otelnet's OtelCORE Notify enables (www.otelnet.com) “intelligent push” wireless filtering services. Subscribers receive personalized, relevant notifications based on their profiles, as well as location and/or availability. These notifications may be in the form of short messages, e-mail or other formats, depending on the subscriber's device and delivery preferences. Notification services include stock portfolio, quotes, price changes; headline news; flight delays and seat availability; and entertainment services. The service also filters content based on subscriber data. Subscribers dynamically can set, change and authorize their availability and permissions, ensuring privacy.

Next-Gen Messaging

Eric Horvitz, senior researcher and manager, Microsoft Research, Adaptive Systems and Interaction Group (research.microsoft.com) is working on a notification platform that will screen out critical messages intelligently from the myriad of information we receive each day. The focus is on modeling users in their context using such information as vision, analysis and activity, as well as considering the value and time criticality of various information sources that users may need. The technology looks at incoming phone calls, e-mails or instant messages and understands their time criticality and relevance to the user, depending on the sender and content.

The idea is to mesh smart assessment of content with intuition about the value of information to manage a user's workload and provide the best information in the least disruptive way, Horvitz says.

Thinmail (www.thinmail.com) provides wireless users with “attachment intelligence,” or an intelligent electronic assistant capable of tasks such as translating documents into simple text and diverting bulky attachments to fax machines.

The service offers full business e-mail capabilities on any wireless device. It's a universal e-mail relay providing intelligent-assistant services to users.

Mail sent through Thinmail's servers is processed by a system providing formatting and filtering services to wireless users. The solution turns e-mail from “push” to “pull” by converting all attachments into private links that act as virtual attachments on any text-enabled device, says Jordan Pollack, Thinmail CEO.

In the future, wireless devices and apps will be embedded with the ability to understand and manipulate language, make sense of what can be seen, heard or felt, as well as find useful patterns in data, draw conclusions based on rules and experience, and respond — without human intervention.

To address consumer fear of the new technology, privacy and security standards must be established. A user may appreciate his phone reminding him to pick up a cake for his spouse's birthday, but may hate a device that inundates him with cake-mix sales. Subscribers must be able to control device intelligence by turning it on or off.

Belk says that with intelligence comes responsibility.

“The capabilities will be there but must be defined in such a manner that it's the user who defines who gets what information,” he says.