February 14, 2000 Web posted at: 8:57 a.m. EST (1357 GMT) by Ephraim Schwartz

From...

(IDG) -- Whether you want to optimize your investments in petrochemicals or save $1.50 on a best seller, you'll probably soon be doing it with a bot.

Short for robot, "bots" -- also called autonomous agents -- are small pieces of code sent out like carrier pigeons across cyberspace to bring back information, such as the best price on a product. Sometimes they go no further than the corporate knowledge base, where they are used as the first response to a help desk or call center request before the call is escalated to a live agent. By automating often time-consuming tasks, e-businesses can improve customer service by giving more consumers what they need much faster than with human intervention.

Bot builder Daniel Sapir, the COO of Boston-based Artificial Life, knows all about how companies can save money while improving CRM (customer relationship management) via bots.

	MESSAGE BOARD
Managing the Net

"It's [a] tremendous cost reduction for CRM," Sapir says.

Sapir quotes Forrester Research statistics that say a typical customer call handled by a live agent costs about $33. The same call handled in a chat-room environment is $7.80. But if a bot can find the answer from the company knowledge base and automatically deliver it to the caller on-screen, the cost is $1.17.

MORE COMPUTING INTELLIGENCE

IDG.net home page

The new bidding war

To bot or not to bot

A mechanical helping hand

Reviews & in-depth info at IDG.net

E-BusinessWorld

Year 2000 World

Questions about computers? Let IDG.net's editors help you

Subscribe to IDG.net's free daily newsletter for IT leaders

Search IDG.net in 12 languages

News Radio

Fusion audio primers

Computerworld Minute

"Once you build your knowledge base, bots automate the process, and it is always there for the customer," Sapir says.

Shopping bot Web site mySimon.com is a typical example of the most popular implementation of bot technology on the Web: getting the lowest price.

At its deepest level, a shopping bot is a piece of software that mimics human behavior, according to John McDowel, vice president of engineering at mySimon Inc., in Palo Alto,Calif.

To get "smart," mySimon.com uses Simon Product Intelligent [SPI] agents, who are real people, to shop online at thousands of e-tailer sites. The SPIs are equipped with a tool called the virtual-learning-agent system, which mimics the shopper's behavior. The result: If a shopper is looking for a sweater, for example, the system will bring back results from a half dozen or so sites that will include not only price but also shipping information, color, and size -- all the typical pieces of information a user would have been looking for.

But others are stretching the potential of bots with enhancements to the searching function: Sapir's company uses bots with natural language comprehension technology to carry on a limited but functional conversation between a customer and the bot, represented on-screen as a character, or avatar, as it is referred to in the industry.

Liechtenstein Global Trust uses bot technology from Artificial Life for its "high net worth" customers, Sapir says. The financial services company, in Liechtenstein, uses the bot to go to predetermined financial Web sites such as those of the Morning Star and The Wall Street Journal, scan data on user-determined criteria, and not only come back with information but also suggest possible actions based on the information.

"For example, if the interest rates are going up, it may be it is time to buy bonds and sell stocks," says Sapir, who adds that the bots are "trained" to spot trends.

Net-tissimo.com, a lifestyle Web site in Egerkingen, Switzerland, launching in the United States, sells wines, electronics, and computer peripherals and also provides travel information. It uses Artificial Life bot technology for its virtual butler that, if asked, can tell a visitor what to eat with a Merlot wine, for example.

"We started by offering information about the products. Then we collected information from our log files to see which questions were too complex for the butler to answer," says Roland Berger, CEO of Net-tissimo.com.

"The butler learns. He doesn't give canned answers. He picks and chooses words and puts them together into a [coherent] sentence," Berger says.

According to Berger, the look-to-buy ratio on the Net-tissimo site has gone from an average of 2 percent to 4 percent pre-bot to a 10 percent sales rate post-bot.

Looking ahead, bots will automate mundane tasks to complex searches.

"In order for an economy to be productive, it has to be automated, or it will remain primitive," Sapir says. "The promise and potential of the Internet is huge, but unless you begin automating things, it will never fulfill its promise."