Even those of us who cover e-commerce often overlook the reality of connecting the "e" part of the equation to the real world. Webvan and Kozmo.com were great online services, but it was the unavoidable and nonvirtual expenses of moving goods around that did them in. Former Webvan business process guru Mick Mountz said the company lost $20 an order, due to the cost of physical logistics.

In 2003, having left Webvan and moved to the East Coast (to escape Webvan's investors, I would guess), Mountz started Kiva Systems, a robotics company that rethought warehouse logistics for the world of e-commerce. Kiva could have saved Webvan, Mountz believes, it if had been around in 1999.

A Kiva-powered warehouse looks like a jumble to a human, but the software sees order in the location of every good tracked.

Kiva is not new, and we've covered it before, but I find it interesting, since it's one of the few U.S. tech companies to take advantage of the our trade imbalance with Asia without actually manufacturing goods in cost-competitive overseas plants.

To recap, the company makes what looks like giant Roomba robots. These big puckbots can lift 3,000 pounds each. Swarms of them live in pick-and-pack warehouses. They move entire shelving units to the employees packing boxes by scooting under the shelves, lifting them on jacks, and then rolling to workstations. At the workstations, humans (who have the "world's best end effectors," says Mountz), pick items off shelves and pack them into boxes. Then the robots park the shelves out of the way of other shelves in motion.

The control systems in each warehouse monitor orders to make sure shelves with the most popular items stay close to the front of the shop and that items that are usually sold together are on single shelving units. This reduces travel time of goods. As Kiva engineers learn more about how their systems work, they tweak the software to improve efficiency. One recent software upgrade: parking shelving units two shelves deep. If a parked-in shelf is needed, it takes a "buddy bot" to move a shelf facing an aisle out of the way, but this scheme dramatically increases warehouse storage density.

As I said, current trade economics work in Kiva's favor. So much U.S. consumer material arrives here via ports now. At docks, shipments are broken down into still-large shipments that get sent to U.S. distribution centers. From many of those centers, shipments get broken down again, either into store-size loads headed to regional brick-and-mortar retail stores, or into individual orders that contain items combined into boxes for e-commerce delivery to consumers. The long consumer supply chain, and the resultant growth in distribution warehouses, is a shift in U.S. retail, driven by both global economics and the growth of the Web. Kiva is riding both waves, as are Amazon and other e-commerce retailers.

Amazon, of course, began experimenting with methods to improve pick-and-pack efficiency early in its existence. But the Web retailer never bought a solution from Kiva Systems--at least not directly. As it happens, two companies Amazon recently decided to buy--Zappos.com and Diapers.com--have been Kiva users. Through those buys, Amazon now owns four Kiva-powered warehouses (Diapers.com has three, Zappos one).

Mountz said Kiva's systems have replaced traditional serial pick-and-pack logistics, such as conveyor belt systems, or warehouses where people move around with totes, with a parallel and fault-tolerant system that takes advantage of the efficiency of a lot of automated and cheap-to-maintain small moving parts (its robots) to overcome the inherent inefficiency of moving 3,000 pounds of shelving and unwanted products around, when all a customer wants is one 6-ounce tube of toothpaste.