Amazon’s AI Strategy is a Thing of Beauty

Tony Chen

7 years ago

You know that feeling when you’ve just seen the most perfect sunset? Or when you watch Roger Federer hit a perfect drop shot in the most unexpected situation? Or that first time you heard “Alexander Hamilton?”

Utter. Beauty.

That’s exactly how I feel as I learn about Amazon’s last-to-leader ascension with AI. As it turns out, AI is the latest and greatest rocket fuel feeding possibly the world’s greatest growth engine that is build upon this principle: Amazon fanatically listens to what the customer wants and gives it to them. This customer obsession compels Amazon to innovate. Those innovations win customers over. Amazon then packages each innovation into a service so that other businesses can win their customers over, too. Here are four glimpses into their story:

Amazon’s laser focus on customers is seen in its full glory in this now-infamous document template. Amazon leaders pitch Bezos on new ideas with “the six-pager” that includes a press release. A press release! New features and innovation are awesome, but how will it feel for the customer? What a clever way to keep a loud customer voice inside the heads of brilliant engineers who spend their days thinking about far-field technology and neural networks.
AI at Amazon rocketed from benchwarmer to superstar using an upside-down recruiting approach. All the other big techs were using the traditional basic research approach, “let’s develop this technology and see how we can use it.” Amazon unabashedly starts with customers, “this is how customers want to use it, let’s develop the technology to work that way.” Those press releases of visionary products represent an all-too-juicy buffet of challenging, technical puzzles to solve for rising AI stars in natural language processing.
Amazon leveraged its best assets to create Alexa. What other company could possibly combine disparate assets like lighting fast data centers (so your “conversation” with Alexa is void of awkward silences), unparalleled cloud service, hardware chops, data wizardry, and a massive, loyal customer base? This is like using your already-perfectly positioned chess pieces to place more chess pieces on the board. Now with 70%+ market share, Amazon’s Alexa is likely collecting at least 2-3x more data than anyone else, enabling Alexa to learn faster and further widen the gap.
Amazon has a pattern of creating something valuable it needs itself and then selling it to the world. Amazon created the best e-commerce platform in the world to sell stuff, but now 50%+ of all e-commerce sales are being sold by third parties. Amazon leads in e-commerce largely due to its ability to quickly crunch data to maximize sales. Today, that capability of data storing, crunching, securing, optimizing (i.e. Amazon Web Services) is utilized by millions of customers — to the tune of ~$5B in revenues in Q42017. And yes, you guessed it: as Amazon built AI to make Amazon.com and Alexa better, it’s no surprise that it started selling AI-as-a-service in 2015. Just as WordPress democratized publishing for the masses, Amazon’s new machine learning offering (called SageMaker) aims to open up machine learning to the software developer masses.

Imagine one day when you could feed your [fill in the blank] data set through SageMaker, ask a few questions through Alexa, and out pops the answer that transforms your business? Your health? Your investments? Your marriage?

Every single minute of every day, the customer is actually telling us what he wants. Who’s really listening? Amazon’s massive success to date is largely due to being one of the most attentive listeners on the planet, recording, analyzing (now with AI), and learning from millions of daily customer interactions.