Amazon has introduced three AI-powered technologies designed to enhance parcel packaging, delivery, and supply chain forecasting, as part of its efforts to streamline operations and improve customer service.
The first of these, Wellspring, is a generative AI mapping tool that pinpoints entrances, car parks, and individual units within apartment complexes to increase delivery accuracy. The system integrates data from multiple sources, including satellite imagery, road networks, building footprints, customer instructions, previous delivery records, and street images, to refine delivery solutions for millions of locations.
Additionally, Wellspring identifies building entrances and post office locations by analysing proof-of-delivery photos and historical delivery data. During trials in 2024, the technology mapped over 2.8 million flat addresses across more than 14,000 complexes and identified optimal parking spots for 4 million addresses.
Amazon has also developed an AI-driven demand forecasting model to predict fluctuations in supply chain demand. By analysing regional sales trends, the system tailors inventory distribution to meet varying global needs more efficiently. The company reports that this model has improved long-term national forecasts for commercial events by 10 per cent and regional forecasts for millions of popular products by 20 per cent, boosting productivity while reducing the carbon footprint of its logistics network.
Lastly, Amazon Robotics is integrating agentic AI capabilities into its robotic systems, enhancing their ability to process natural language, reason, and operate autonomously. Using Vision Language Models (VLMs) and action-guiding policies, robots can now follow plain-text instructions.
This advancement will upgrade systems like Proteus—an autonomous mobile robot that transports customer orders—into adaptable assistants capable of manoeuvring heavy items in confined spaces. Amazon stated that this innovation will allow employees to focus on higher-level tasks requiring critical thinking and problem-solving.
Recent Stories