FORTUNE — There are few dot-com era startups more ignominious than Webvan, the online grocery business that spent over $800 million in three years before collapsing in 2001. It remains a cautionary tale for entrepreneurs like 26-year-old Toronto-raised Apoorva Mehta, whose seven-month-old business Instacart is trying to succeed in a $568 billion domestic market in which Webvan notoriously failed. In doing so, he finds himself going toe-to-toe with Amazon (AMZN) — not to mention a swath of other competitors — where he once worked as a software design engineer, creating complex strings of math to find the most efficient routes for shipping packages to customers.
“They were delivering your groceries along with Xboxes,” recalls Mehta, who collaborated with AmazonFresh employees. Delivering groceries alongside traditionally higher-margin items like electronics was done to help temper the low margins and pricey overhead typically associated with procuring, storing, and shipping produce. “That may seem like an interesting idea, but how many times do you need to buy an Xbox?”
With $2.5 million from startup incubator Y Combinator, Khosla Ventures, and others, Mehta’s startup approaches online groceries differently from AmazonFresh, which expanded beyond Seattle to Los Angeles earlier this month and could reach the San Francisco Bay Area later this year. AmazonFresh’s expansion will reportedly rely on the construction of new warehouses near cities. Instacart uses an already existing infrastructure, one that includes well-known supermarkets and the hustle of some 200 contracted personal shoppers across the San Francisco Bay Area, where Instacart is available.
MORE: Why Apple’s new MacBook Air is the ultimate road warrior’s notebook
Here’s how it works: Customers shop online or via mobile app, choosing from the inventories of chains like Safeway (SWY), Whole Foods (WFM), Trader Joe’s, and Costco (COST), down to small, independent markets. Each order may have items from several stores. Orders with 15 items or less are eligible for one-hour delivery. Otherwise, two-hour, same-day, or a delivery at a later date and time are also options. Instacart charges a small premium based on the purchase size.
Once a customer places an order, a smartphone app notifies a personal shopper. It lets them know which store to go to, and which aisle and shelf the item is located in. To maximize efficiency, each shopper works on filling several orders at once, the equivalent of 60 or 70 items. (If a customer orders from three different stores, three personal shoppers are assigned and rendezvous afterwards to merge their purchases into one delivery.) The same app also helps shoppers deliver, suggesting routes that factor in traffic, weather, sports games, and city construction. The result: Customers get their purchases in as little as an hour.
Of all the businesses he’s helped fund, Instacart remains one of the most used by Y Combinator co-founder Graham and his wife Jessica. “Instacart is one of those rare products that’s surprisingly great,” he says. “You don’t realize how good such a thing could be till you try it.” Perhaps. But Instacart has a long way to go before it achieves the reach of FreshDirect, the 11-year-old online service serving Manhattan, Philadelphia, and New Jersey. But the startup’s low-cost model may lend itself better to rapid expansion. Available in the Bay Area now, Mehta wants to be in 10 major metropolitan areas by the end of next year.
MORE: The Apple e-books trial takes a detour into the absurd
One issue he may have to eventually deal with is the stores, with which Instacart has few official ties. (Mehta suggests several partnerships are in the works.) Instacart’s ability to let users browse competitors’ inventories to mix-and-match items likely won’t make some chains happy. And the new “Instacart Plus” feature launched earlier this month pushes comparison shopping even further, tracking every store in its database and serving up the cheapest price for available items like that jar of peanut butter. “We may end up cannibalizing sales from stores like Safeway, but in the end, this is the best decision for the customers,” says Mehta. That’s a retail philosophy even his former CEO, Jeff Bezos, might agree with.
