← All topics

#Amazon

9 stories · 5 issues · daily Amazon news timeline


Coverage & briefs

Updated with each publication

Reuters says Amazon is planning a World Product Graph to index products across the web

According to Reuters, Amazon has an internal plan for a “World Product Graph” that would collect data on products sold anywhere online, including items not sold on Amazon. The goal is to keep shoppers and AI agents inside Amazon by making its product index more useful than rivals such as Google or Reddit. The document also says the extra data could improve Alexa for Shopping, formerly Rufus, which it describes as inaccurate and costly to run. Amazon told Reuters the draft may not reflect current plans.

AI-generated

AWS adds Data Capture to SageMaker HyperPod for live inference logging and retraining

AWS says SageMaker HyperPod now includes Data Capture, a built-in feature that records live inference inputs and outputs to Amazon S3 with minimal latency impact and no effect on model availability. The post walks through one HyperPod cluster where a home-appliance support router captured production traffic, Amazon Athena identified a new-product blind spot, and a SQL policy labeled the data for fine-tuning. AWS says a 60-second LoRA run raised recall on the new category from 0% to full recall without regression on existing classes.

AI-generated

According to Ars Technica AI, Amazon unveils Ring Robotic security drone and rover for homes and businesses

According to Ars Technica AI, Amazon announced two new Ring Robotic devices on November 10: the Ring Rover Cam and Ring Drone Cam. Both use an AI Pilot system that combines cameras, specialized vision models, on-device sensors and cloud processing to map interiors and navigate around obstacles. Amazon expects the Rover to reach customers at the end of 2026 and the Drone in early 2027, pending FCC approval for the flying model.

AI-generated

AWS adds single-shot RAG and retrieval-only modes to Amazon Bedrock Managed Knowledge Base

AWS Machine Learning said on its official blog that Amazon Bedrock Managed Knowledge Base now supports single-shot RAG and retrieval-only requests through AgenticRetrieveStream. Single-shot RAG retrieves once and lets a foundation model generate the answer directly, which suits simple lookups, while agentic retrieval can run multiple searches for more complex questions. Developers can also set maxAgentIteration to 0 and disable generation to return deduplicated source passages, making it easier to review policy text directly. Existing requests with values from 2 to 5 remain unchanged.

AI-generated

AWS adds NVIDIA MIG GPU partitioning to SageMaker Inference

AWS announced on its official blog that SageMaker Inference now supports NVIDIA Multi-Instance GPU partitioning. The feature lets a physical GPU be divided into isolated partitions with dedicated compute and memory, so more models can share one endpoint and teams can choose a better-sized slice instead of a whole accelerator. AWS said the capability works with inference components, while single-model, serverless, and heterogeneous instance pool endpoints are not supported. Salesforce has helped define requirements and validate the initial implementation.

AI-generated

The Verge AI reports Amazon is letting shoppers text Alexa to place orders

According to The Verge AI, Amazon is adding a text-based shopping feature to Alexa for Shopping that lets customers message the assistant with a product description or photo and have it search Amazon listings, suggest matches, and place the order. The contact appears as “Amazon Alexa,” and the feature can handle groceries, gift ideas, similar-item searches, and price tracking based on user preferences. Amazon says it is rolling out first to a small group of US customers, then to millions more in the coming weeks.

AI-generated

AWS makes Web Answer generally available on Amazon Bedrock AgentCore

AWS Machine Learning announced the general availability of Web Answer on Amazon Bedrock AgentCore. The tool lets agent builders ask a natural-language question and receive a synthesized, source-cited answer through a single call. AWS says Web Answer handles search, fact-checking, synthesis, and attribution for current web sources, so developers do not need to build their own retrieval pipeline. It is available through AgentCore Gateway and can be added as a managed connector.

AI-generated

AWS Machine Learning describes a VLM-based quality system for Amazon A+ Content

In an AWS Machine Learning blog post, the company described a pipeline for evaluating Amazon A+ Content by rendering each product page as a single snapshot and scoring it with a vision-language model. Those semantic scores are combined with OCR readability signals in a CatBoost classifier trained on expert annotations. SHAP then explains the decisions and produces actionable recommendations for selling partners. On 5,478 pages across 10 categories, the fused model reached 0.94 accuracy and 0.95 F1, and revised pages improved by 1.4 points on a 10-point scale.

AI-generated
// related topics — frequently covered alongside #Amazon