Amazon is betting on AI to improve its operations and help customers with new models.
During the AWS Re:Invent event, CEO Andy Jassy talked about how the company is using AI to improve operations and what’s next on the AI landscape.
Amazon is using AI to have better recommendations and personalization in the retail business and to optimize fulfillment center operations. “We hope to deliver items to you in less than an hour in a couple of years,” Jassy said.
Amazon has 1,000 different buildings or nodes and in Latin America the company has several fulfillment centers in Mexico and Brazil.
“Everything we do is optimized to get the right product in a fulfillment center or building close to the end customer to save on transportation time, which means we get it to you faster and we do it for a lower cost,” Jassy said.
According to the CEO, the use of AI to manage inventory meant an improvement in long-term demand forecasting accuracy of 10%. “We’ve also improved regional prediction accuracy by over 20%,” he added.
Amazon has over 750,000 robots operating in various fulfillment centers, with all sorts of AI in them. At the fulfillment center launched recently in Louisiana, the company has seen 25% faster processing time because of the combination of AI inventions and robotics. “We believe we’re going to have 15% lower costs to serve during the holidays,” Jassy forecast.
Customer experience
The US company is also using genAI to improve the customer service chatbot and shopping experience.
Amazon launched Rufus, a shopping genAI agent that can make recommendations based on customer information.
The company is also re-architecting the ‘brains’ of virtual assistant Alexa with multiple foundation models. “It’s going to not only help Alexa answer your questions even better but understand and anticipate your needs and actually take action,” Jessy said.
Also, with the information collected about the user, such as what has been bought and where they live, Amazon can recommend the nearest physical location to return a product or predict if the user is frustrated and needs to be connected to a human for a resolution.
According to Jassy, since the company added the generative AI ‘brain’, customer satisfaction has improved.
Amazon is also using AI to make it easier for sellers to complete the forms to add products to their e-commerce offering. The company also built a branding tool using genAI so sellers can have their own information page. Currently, about 500,000 sellers are using the tool.
New AI models
Amazon launched Amazon Nova, a new state-of-the-art aggregation model for computer intelligence.
Nova is a set of six models, including text and multimodal models with text, image or video.
Four of them are available today via AWS’s platform Bedrock while the speech-to-speech and any-to-any multimodal models will be available in 1Q25.
Multimodal models can input any kind of format (text, image or video) and output any format.
Amazon also announced Amazon Nova Canvas, a generation model that allows for the creation of images using natural languages and Amazon Nova Reels, which can create videos.
According to AWS, Nova is about 75% less expensive than the other leading Bedrock models.
link
