hen programmers first started to work on filters and agents, some joked about the
consequences. My agent would call your agent and negotiate everything from the price of my
products to lunch arrangements. A few thinkers warned that big economic changes could happen
when the techniques matured.Intelligent agents, which could not only search but act on the results, could go off crazed in the middle of the night, reacting to news or competitive threats. There's no telling how prices could rise or fall.
Some experts now caution that enormous spontaneous price wars could erupt when shopping agents start to review their competitors' activities. The fundamental technology for this is already in place.
Today, companies with pioneering search-agent technology can put a dent in consumer purchase prices. While the technology is still in its earliest days, you can see the power and promise of online shopping agents.
For example, my team at Currid & Co. recently visited several sites to look for the price of an Iomega Buz. Though there were differences in the results found by each agent, most found up to 23% price differences, with a low price of $153 and a high price of $199.
The next step will come when the agents get a little smarter. When agents search each other, they will make quick comparisons. Before you know it, agents in some areas will be given permission to reset prices based on market conditions. Agents for sellers (providers), brokers, and buyers will work-or fight-for business.
Researchers studying this issue include Jeffrey Kephart from the IBM Thomas J. Watson Research Center. He recently constructed a model to predict what might happen in an advanced electronic market. He found that agents for the broker and consumer raged with devastating price wars. Profits were constantly jeopardized as agents fought for business. But smart buying power doesn't necessarily put the customer in control. Agents for sellers will ferret out unprofitable customers.
Another researcher, Alexander Chislenko of MIT, built a different model but ended up with similar results. He concluded that economies run by agents verged on chaos and said controlling them "would be like trying to control a car that was traveling at 500 miles an hour-not easy." You can find details of his work on the Web at www.lucifer.com/~sasha/HEDG/HEDG.html.
What does this mean to corporate IT? Get the radar up. Though I doubt we will see wholesale changes for several years, agent technology is too close to go untested. Like other megatransitions-universal price coding, scanners, online transactions-agent technology will be used on a large scale, industry by industry. But small or focused attempts could happen with technology available today.
Consider the example my team used. Using the same technique, a reseller or competitor of removable storage units could perform a distribution, price, and market survey in a few hours. Decisions, made today by humans, can be made faster tomorrow to meet or beat the competition. Targeted marketing programs, available in price competitive areas, could also be set, making a company very nimble.
In the soon-to-be new world, humans won't be needed to intervene and apply judgment. The agents will be able to gather up information and decide for themselves. Realistically, businesses would want to give their agents a spending limit-such as "don't reduce prices more than 10% without a [human] manager's approval." Rules and controls could limit the risk of spontaneous price wars.
No doubt, agent technology will come to every industry. Business managers and IT professionals need to collaborate on the issue of agent control. It's a great time to experiment with models and pilots because there's no competitive pressure. When the technology matures, you will be prepared to turn over the keys of control to your computer.
Cheryl Currid is president of Currid & Co., a technology consulting firm in Houston. She can be contacted at When programmers first started to work on filters and agents, some joked about the consequences. My agent would call your agent and negotiate everything from the price of my products to lunch arrangements. A few thinkers warned that big economic changes could happen when the techniques matured.
Intelligent agents, which could not only search but act on the results, could go off crazed in the middle of the night, reacting to news or competitive threats. There's no telling how prices could rise or fall.
Some experts now caution that enormous spontaneous price wars could erupt when shopping agents start to review their competitors' activities. The fundamental technology for this is already in place.
Today, companies with pioneering search-agent technology can put a dent in consumer purchase prices. While the technology is still in its earliest days, you can see the power and promise of online shopping agents.
For example, my team at Currid & Co. recently visited several sites to look for the price of an Iomega Buz. Though there were differences in the results found by each agent, most found up to 23% price differences, with a low price of $153 and a high price of $199.
The next step will come when the agents get a little smarter. When agents search each other, they will make quick comparisons. Before you know it, agents in some areas will be given permission to reset prices based on market conditions. Agents for sellers (providers), brokers, and buyers will work-or fight-for business.
Researchers studying this issue include Jeffrey Kephart from the IBM Thomas J. Watson Research Center. He recently constructed a model to predict what might happen in an advanced electronic market. He found that agents for the broker and consumer raged with devastating price wars. Profits were constantly jeopardized as agents fought for business. But smart buying power doesn't necessarily put the customer in control. Agents for sellers will ferret out unprofitable customers.
Another researcher, Alexander Chislenko of MIT, built a different model but ended up with similar results. He concluded that economies run by agents verged on chaos and said controlling them "would be like trying to control a car that was traveling at 500 miles an hour-not easy." You can find details of his work on the Web at www.lucifer.com/~sasha/HEDG/HEDG.html.
What does this mean to corporate IT? Get the radar up. Though I doubt we will see wholesale changes for several years, agent technology is too close to go untested. Like other megatransitions-universal price coding, scanners, online transactions-agent technology will be used on a large scale, industry by industry. But small or focused attempts could happen with technology available today.
Consider the example my team used. Using the same technique, a reseller or competitor of removable storage units could perform a distribution, price, and market survey in a few hours. Decisions, made today by humans, can be made faster tomorrow to meet or beat the competition. Targeted marketing programs, available in price competitive areas, could also be set, making a company very nimble.
In the soon-to-be new world, humans won't be needed to intervene and apply judgment. The agents will be able to gather up information and decide for themselves. Realistically, businesses would want to give their agents a spending limit-such as "don't reduce prices more than 10% without a [human] manager's approval." Rules and controls could limit the risk of spontaneous price wars.
No doubt, agent technology will come to every industry. Business managers and IT professionals need to collaborate on the issue of agent control. It's a great time to experiment with models and pilots because there's no competitive pressure. When the technology matures, you will be prepared to turn over the keys of control to your computer.
Cheryl Currid is president of Currid & Co., a technology consulting firm in Houston. She can
be contacted at cheryl@currid.com.
Back to Columnists
Send Us Your Feedback
Top of the Page
University of San Diego seeking System Administrator 2 in San Diego, CA
Hebrew Senior Life seeking Network Analyst in Boston, MA
Cirrus Design seeking Web Architect in Duluth, MN
Comcast seeking Tier 4 CRAN Network Engineer in Chelmsford, MA
Lowe's seeking Network Engineer II in Mooresville, NC
For more great jobs, career-related news, features and services, please visit our Career Center.