Bot Till You Drop

By Chris Taylor

When you've garnered a net worth of $7 billion, a 12-million-strong customer base and the success story of the '90s, you'd think you could afford to act mean to the other guy once in a while. Not Jeff Bezos. Rather than leaning Gates-like on his competitors, Amazon.com's entrepreneur extraordinaire last week issued them all invites to his party.

For one thing, you can now set up your own little e-commerce homestead on a corner of Amazon real estate called zShops. But much more important, Amazon's main search tool--that little text box in the upper-left-hand corner of the home page--has started pointing customers toward sites other than its own. "Think of a small-town merchant who suggests you might try someone else down the street," says Bezos, with his trademark earsplitting laugh.

Nice guy, huh? Nah. Killing the competition with kindness is more like it. For Bezos has long realized what Wall Street seems to have just awakened to: you can't stick around on the profit margins of a bookstore, or even a book-CD-video-toys-electronic store with the odd auction thrown in (which last year had sales of $1 billion and zero earnings as usual). To stay ahead of the curve, Amazon needs to build the ultimate e-commerce portal: the holy grail; the must-have home page for all online shoppers. And for that you need a shopping agent.

Shopping agents, better known as bots, are practically prehistoric in Internet terms. They've been around in one form or another for four years. Their basic purpose is to search the Web and compare all the prices you can possibly pay for the same item, making you an expert bargain hunter in a single click. Once considered a threat to e-business--after all, sellers benefit when the consumer has incomplete knowledge of prices--bots are now on the buying list of every major player in the Internet industry. They're into their second generation and much improved. And they have gone mainstream.

"At first, store owners were very concerned that if their prices weren't low, people wouldn't buy from them," says Marcus Zillman, who tracks and lists electronic agents at botspot.com "That hasn't happened. Mainly, the shopping bot allows people to focus on what they're looking for."

How much do the big guys want this technology? Check out the balance sheet. Firefly, one of the earliest bots, was swallowed by Microsoft for an undisclosed sum last year, as was CompareNet.com six months ago. Excite snapped up Jango.com for $35 million in stock. Lycos employs the services of a bot start-up called Frictionless.

And just over a year ago, Amazon shelled out $180 million for a price-comparison site called junglee.com They tinkered around with its virtual database technology until relaunching it last week as the more prosaically named All Product Search. Wall Street was buying, sending Amazon's stock up 23%, to $80.75, in a day.

But the trouble with bots that get sold to big corporations, not surprisingly, is that they tend to get a little compromised. In junglee's case, the ability to compare prices at other book sites seems to have disappeared. Type in Tom Clancy in Amazon's All Product Search, and there's no danger of your buying Rainbow Six from Barnes and Noble or Borders.

How can Bezos get away with that? Because he believes most nonproprietary shopping agents aren't yet ready for prime time. But there's a whole host of cool new bot technologies out there that may force Amazon to think again. Take mySimon.com the most successful of the independent bot sites. It's about to start a service that will e-mail you whenever any online merchant lowers its price for a particular item to an amount you're willing to pay. Feeling frugal about that $150 pair of binoculars? No problem; sit back and wait until the market takes it down to $99. Then jump on it. (As you would expect, there are auction bots that do exactly the same for eBay and its clones.)

Pretty neat, you think? Just wait until the third generation of shopping agents moves out of the lab. Even now, folks at M.I.T. and IBM are preparing for a world in which every transaction becomes a complex trade deal between a pricing bot acting for the site and a shopping bot acting for you. "Dynamic pricing, that's the big notion," says Professor Pattie Maes, director of the software-agents group at the M.I.T. media lab. "After all, fixed prices have been around only for a couple of hundred years."

What Maes predicts, you had better believe. After all, she was the pioneer of the earliest shopping-bot technology known as collaborative filtering, which became popularized as Amazon's ubiquitous "people who liked this product also bought X" link. Now she's helping to bring us an e-commerce world based on the Priceline.com model--in other words, every seller will act like an airline with constantly changing pricing and the ability to negotiate. It's auto haggling.

This bot-driven universe won't arrive until a few kinks have been ironed out. Right now, says Jeff Kephart, manager of the agents and e-merchant phenomena group at IBM, price bots don't understand that undercutting your competitor is not always smart. "This gives rise to price wars," says Kephart, who in tests has watched the sell bots give the store away in a competitive frenzy. "They're pretty dumb," he notes. "We have to give them a sense of anticipation."

Which means that Bezos can breathe easily for the moment. However, most up-and-coming bots have just the sort of customizable features that made his site such a hit. You will, for instance, be able to tell your bot to transact only with companies that offer two-year-warranties or free shipping or that don't run sweatshops in Malaysia.

Still, more than most, Bezos is likely to adapt to this brave new bot world. He understands its fundamental creed. "The balance of power online fundamentally shifts from merchants to consumers," he says. Indeed, thanks to shopping bots, the party won't just be at his place anymore.