Amazon is acquiring Kiva Systems for $775 million in cash. Kiva makes robotic systems for picking, packing, and shipping products in fulfillment centers for distribution operations. It seems clear that Kiva is intended to play a dual role at Amazon: supporting Amazon’s own distribution centers, and generating expanded revenues through the sale of Kiva systems to other companies.
There is a parallel with what Amazon has been doing in cloud services: Amazon developed an extensive set of capabilities for data center operations, which it needed to support its massive e-commerce business, and several years ago began selling these capabilities to other companies as well as using them internally. Amazon Elastic Compute Cloud has now become a leading provider, perhaps the dominant provider, in the cloud services marketplace.
Use of a technology investment both to support internal operations of a company and as the base for an externally-saleable product or service has a strong appeal; however, it can be fraught with problems. Priority decisions in product development are likely to become highly politicized due to the conflicts between internal needs and the demands of the external marketplace, and potential external customers can be scared off by fear of being put in the position of competing with their supplier. Amazon’s success with Cloud, however, builds confidence in their ability to navigate these tricky waters successfully.
Interestingly, Kiva is backed by Bain Capital Ventures.
I read some year ago of Amazon’s current way of retrieving an ordered product.
Someone on roller skates goes and gets it.
Perhaps they have changed in the interim but the author at the time said one would be surprised at their low-tech means of getting the article to ship.
There is some pretty neat stuff out there.
Our car club was giving a tour of our dealer’s parts dept – and they have an automated thing – like a wine dumbwaiter – to bring up the part – open the door once the part has been keyed on the computer.
There’s some interesting stuff out there.
I wonder if Kiva’s new system will have robots reading bar codes?
There have of course been quite a few cases where robots were installed and resulted in higher costs and lower flexibility than could have been achieved with with streamlined manual methods….Roger Smith’s roboticization of GM assembly plants being the classic example.
Related post/discussion at Lean Blog: automation is not always the answer
This product description page suggests that Kiva has put a lot of emphasis on flexibility and that its systems are human/automation combinations rather than an attempt at complete roboticization.
Some interesting discussion at the first link…note especially the comment by Tom Benson about geographical/cultural factors in innovation:
“Mick said in several interviews that he tried to sell his robotics concept in Siilicon Valley originally, but the VCs here wouldn’t take him seriously. He finally went to Boston, where there is a more “business model” oriented robotics culture.
In our robotics startup several years ago (Readybot) we saw the same problem. Bay area people just don’t see robotics as a viable business; they really see it as a giant science fair project.”
Interesting article on automation David. I have always thought that those ‘self serve” units are more trouble than they are worth. They have to have an attendant standing by when a customer has problems.
And lately, in supermarkets the state of California has all but abolished them because of liquor sales by underage minors.
So I suppose if you want to keep one open you have to have someone standing by verifying via ID that a customer can buy liquor.
In vehicle assembly, I wonder what GM – and Roger Smith – did wrong in automation that the Japanese are doing right?
I do know that with Mercedes-Benz automation has revolutionized their production.
Part of the reason they have been so expensive in the past is the army of factory workers – assembling, inspecting, taking apart problem parts and reassembling….
Now it is almost all automated. It’s done right the first time with far fewer people in the assembly process.
An E Class today is about the same cost in dollars as it was in the early 90s, thanks to automation.
One problem with supermarket self-checkout systems is that so many of them are poorly designed from a customer-interface standpoint. The ones at Giant are especially awful, IMO…they’re like a science fair project that the kid didn’t have time to finish up properly because the deadline caught him unawares.
Even with good design, though, there’s a fundamental problem…for non-bar-coded items like fruits and vegetables, it’s much more labor-intensive for a customer to use the menu hierarchy to find the bar code for, say, bananas than for an experienced checker who knows many of the codes by heart.
David – just speaking as a consumer – I prefer to have a bit of conversation at my checkout – plus those things almost always give you “some problem”. It is so bad that they have a person whose sole job is to get people out of trouble with these automated terminals. In order for me to use one of those (a) the human checkout lines have to have some large lines or (b) they have to be all closed. I use these automated units only reluctantly.
As a programmer I am always amazed that the credit card – or debit card terminals don’t have a uniformity of use to them – how many times have you been detained a bit longer because it wanted one more “Yes” for you to acknowledge?
programs involving interaction should be intuitive – they should respond the way a typical person would act given the incoming data.
Video of a Kiva system in action.
At first glance, at least, it seems to create a good interaction of humans and machines. The robots do the grunt labor and tedious inventory tracking, and the people do things that require manual dexterity and provide an implicit sanity check on the items pulled.