Amazon is set to deploy an artificial intelligence tool aimed at assisting third-party sellers in swiftly resolving account issues and accessing sales and inventory data. The company announced on Thursday that the product, named Amelia, is launching in beta for selected U.S. sellers, with broader availability planned for later this year. Described as an “all-in-one, generative-AI based selling expert,” Amelia will be accessible via Seller Central, Amazon’s internal dashboard for third-party merchants.
Amelia represents the latest in a series of generative AI tools that Amazon has introduced over the past year, building on the momentum generated by OpenAI’s ChatGPT. Other recent AI innovations from Amazon include Rufus, an AI-powered shopping assistant, Q, a business chatbot, and Bedrock, a generative AI service for cloud customers. The company also has plans to enhance its Alexa voice assistant with generative AI features and has invested billions in OpenAI competitor Anthropic, marking its largest venture deal to date.
Amazon’s CEO Andy Jassy highlighted the substantial opportunities presented by generative AI, comparing its potential to transformative technologies such as the internet and cloud computing. In a statement to investors earlier this year, Jassy emphasized the necessity of increased capital spending to leverage this opportunity.
Competitors Google and Microsoft have also released similar generative AI tools, amidst predictions that the market could exceed $1 trillion in revenue within a decade. Amazon’s e-commerce platform has increasingly incorporated AI, featuring AI-generated product review summaries and tools for third-party sellers to create listings and ad photos.
At Amazon’s annual seller conference in Seattle, the company unveiled new tools enabling sellers to generate AI-based video ads and bulk-write product listings. Amazon is also employing generative AI for personalized product recommendations and listings tailored to individual shoppers’ histories. For example, descriptions for a box of cereal might include “gluten free” if a shopper typically searches for that term.
Third-party sellers have been integral to Amazon’s e-commerce operations, accounting for at least half of all goods sold on the platform since 2017, with this figure rising to 61% in the second quarter of the current year. Dharmesh Mehta, Amazon’s Vice President of Worldwide Selling Partner Services, noted in an interview that AI service usage among sellers is on the rise, with over 400,000 third-party sellers utilizing the AI listing tool, up from 200,000 in June.
Amelia is projected to address critical issues for third-party merchants, particularly in troubleshooting account problems. Currently, Amazon supports sellers through extensive teams that handle account suspensions, inventory issues, and business growth on the platform. Sellers have historically expressed frustration with delays in resolving account issues and the challenge of reaching human support promptly.
Amelia aims to streamline these processes by investigating account issues and potentially resolving them automatically. Mehta explained that instead of submitting a form for missing inventory, sellers could instruct Amelia to file a claim or resolve the issue autonomously.
Amelia uses Bedrock, a software tool allowing access to large language models from Amazon and other companies like Anthropic and Stability AI. It is trained on public data from the web, alongside Amazon’s seller resources and FAQs but is not trained on seller-specific data, which remains closely protected.
Amazon stated that Amelia employs retrieval-augmented generation (RAG), a prominent AI framework that combines generative AI with traditional information retrieval methods. This approach enables the extraction of specific information from Amazon’s internal systems without storing it or incorporating it into the training data.
