Amazon president and CEO Andy Jassy, in his annual letter to shareholders, detailed how the company is building new capabilities and services, highlighting how generative AI would fit into its operation.
Jassy said in the letter Amazon gives a lot of thought to how the company can optimize those in the organization who create functions and processes that allow Amazon to tackle fresh challenges and continue advancing its capabilities. They go about it, Jassy wrote, by building what Amazon characterizes as “primitive” services. Amazon defines these as discrete, foundational building blocks that builders can weave together in whatever combination they desire.
Jassy noted the 2003 AWS Vision document published by Amazon Web Services defines primitives as raw parts or the most foundational-level building blocks for software developers. Primitives are not solutions to challenges Amazon faces by themselves, he wrote, noting they are meant to be used together to provide maximum developer flexibility.
Jassy explained:
Over the last 20 years, primitives have been at the heart of how we’ve innovated quickly. One of the many advantages to thinking in primitives is speed. Let me give you two counter examples that illustrate this point. First, we built a successful owned-inventory retail business in the early years at Amazon where we bought all our products from publishers, manufacturers and distributors, stored them in our warehouses, and shipped them ourselves. Over time, we realized we could add broader selection and lower prices by allowing third-party sellers to list their offerings next to our own on our highly trafficked search and product detail pages. We’d built several core retail services – such as. payments, search, ordering, browse, item management –that made trying different marketplace concepts simpler than if we didn’t have those components. A good set of primitives? Not really.
It turns out that these core components were too jumbled together and not partitioned right. We learned this the hard way when we partnered with companies like Target in our merchant.com business in the early 2000s. The concept was that target.com would use Amazon’s e-commerce components as the backbone of its website, and then customize however they wished. To enable this arrangement, we had to deliver those components as separable capabilities through application programming interfaces. This decoupling was far more difficult than anticipated because we’d built so many dependencies between these services as Amazon grew so quickly the first few years.
This coupling was further highlighted by a heavyweight mechanism we used to operate called ‘NPI.’ Any new initiative requiring work from multiple internal teams had to be reviewed by this NPI cabal where each team would communicate how many people-weeks their work would take. This bottleneck constrained what we accomplished, frustrated the heck out of us and inspired us to eradicate it by refactoring these e-commerce components into true primitive services with well-documented, stable APIs that enabled our builders to use each other’s services without any coordination tax.
As it was dealing with Target and NPI complications, Amazon was thinking about building a new set of infrastructure technology services that would allow it to move more quickly and external developers to build anything they imagined. This set of services became known as AWS.
“The above experiences convinced us that we should build a set of primitive services that could be composed together how anybody saw fit,” Jassy added. “At that time, most technology offerings were very feature-rich and tried to solve multiple jobs simultaneously. As a result, they often didn’t do any one job that well. Our AWS primitive services were designed from the start to be different. They offered important, highly flexible but focused functionality.”
As it progressed, AWS launched the Amazon Elastic Compute Cloud in August 2006 and Amazon SimpleDB in 2007. Company observers realized Amazon was building a set of primitive infrastructure services that would allow its partners to build anything they could imagine much faster, more cost-effectively, and without having to manage or lay out capital upfront for the data center or hardware.
As AWS unveiled these building blocks over time, Jassy said, whole companies sprang up quickly on top of AWS. Airbnb, Dropbox, Instagram, Pinterest and Stripe were among the businesses that reinvented themselves on AWS, Jassy said. The building blocks developed also helped enable the operation of streaming services such as Netflix, Disney+, Hulu, Max, Fox and Paramount and even supported critical government agencies. One of the lesser-recognized beneficiaries was Amazon’s own consumer businesses, which innovated at dramatic speed across retail, advertising, devices, Prime Video and Music, Amazon Go, drones and other operations, capitalizing on the speed with which AWS let them build. Primitives, done well, quickly accelerate builders’ ability to innovate, Jassy said.
Another example of how Amazon developed its primitives is in logistics. Because the company built an initial set of logistics primitives, Amazon was able to introduce Fulfillment by Amazon in 2006, allowing sellers to use its existing distribution network to store items and then have the company pick, pack, and ship them to customers, with the bonus of these products made available for fast, Prime delivery. As more merchants began to operate their own direct-to-consumer websites, many still wanted to use Amazon fulfillment capabilities while also accessing the company’s payments and identity primitives they could use to drive higher-order conversion on their own websites, as Prime members have already shared this payment and identity information with Amazon. So, Jassy pointed out, a couple of years ago, Amazon launched Buy with Prime to address the customer aspiration. Prime members can check out quickly on DTC websites as they do on Amazon and receive fast shipping speeds on Buy with Prime items, increasing order conversion for merchants by about 25% versus the default experience.
Amazon has built out logistics operations to include new, same-day fulfillment facilities located in the largest metro areas around the United States, with 58 under operation. These facilities house Amazon’s, top-moving 100,000 SKUs, and they also can incorporate millions of other SKUs injected from nearby fulfillment centers into the same-day facilities. Within those operations, the time required to go from picking a customer’s order to being ready to ship has been streamlined to 11 minutes. Same-day fulfillment facilities offer Amazon’s lowest cost to serve in the network, Jassy said, and the experience with them has been so positive for customers that Amazon plans to double their number.
As it advances, Amazon is looking at how to build new capabilities, such as everyday needs, including perishable food such as milk and eggs, to any order customers make as well as expanding drone delivery.
Generative AI is critical to the next generation of primitives and new capabilities, Jassy said.
Jassy said GenAI has become important to Amazon in more ways than most people realize. At the start, primitives can be part of building foundation models that are critical as a basis of AI development. At the next level, Amazon partners can take what has been built and create customized functions using their own proprietary data and leveraging a cloud provider’s security and features to build a GenAI application. In addressing business needs, the application layer is the epitome, and Amazon has developed various GenAI applications — such as Rufus, the company’s new, AI-powered shopping assistant — while offering partners advertising capabilities, as well as customer and seller service productivity apps.
“We’re also building several apps in AWS, including arguably the most compelling early GenAI use case, a coding companion,” Jassy said. “We recently launched Amazon Q, an expert on AWS that writes, debugs, tests and implements code while also doing transformations like moving from an old version of Java to a new one and querying customers’ various data repositories to answer questions, summarize data, carry on a coherent conversation and take action. Q is the most capable work assistant available today and evolving fast.”
