In the late 1990s, Wall Street and the media perceived Amazon as an "unexciting and marginally profitable online retailer," writes Brad Stone in his new book, The Everything Store.

Google was poaching engineers en masse and the search giant had done a number on Amazon's marketing, forcing a competition with eBay over advertising alongside Google results for popular keywords like flat-screen TV. "They were essentially paying a tax to Google on sales that began with a search," writes Stone. And if that weren’t enough, "Internet users were starting their shopping trips on Google, putting an unwelcome intermediary between Jeff Bezos and his customers."

Bezos was fond of saying that the only way out of this predicament was invention, or as he put it, "inventing our way out." But convincing overseers to approve what Stone calls a "series of improbably, expensive, and risky bets" wouldn't be easy. Bezos needed to prove that Amazon was a tech company and not a retailer, as they had thought. 

The solution was simple: Bezos hired the right people and gave them vague titles. 

One such "technologist" was Udi Manber, a computer science professor tasked with "using technology to improve Amazon’s operations and invent new features," says Stone. Perhaps his biggest innovation was a tool called Search Inside the Book, which emulated the physical experience of flipping through a book by letting shoppers view the first few pages of a title. The innovation was so well-received that it even earned a feature story in Wired magazine. 

But Bezos still wasn't convinced that Amazon had made the transition. 

Then, in early 2002, a computer book publisher named Tim O'Reilly arrived in Seattle. O'Reilly showed Bezos something that would enable Amazon to develop a series of online tools called application programming interfaces (APIs). Here, third parties could harvest data about prices, products, and sales rankings, allowing Amazon to parcel out sections of its store, then let other websites build on top of them. "Soon other websites would be able to publish selections from the Amazon catalog," writes Stone, "including prices and detailed product descriptions, and use its payment system and shopping cart." 

Bezos liked the idea and, after several meetings, gave it the go-ahead. The first developer conference was held that spring and, before Bezos knew it, he was giving the new group a formal name, Amazon Web Services (AWS). 

Today, AWS is the backbone of Silicon Valley, a place where start-ups like Pinterest "rent space and cycles on Amazon's computers," says Stone, "and run their operations over the Internet as if the high-powered servers were sitting in the backs of their own offices." The business might seem completely unrelated to retail, and that's exactly the way Bezos wanted it. His push for invention had given Amazon a new way forward.