Amazon unveils three AI-based delivery support solutions
Amazon is taking a multi-pronged approach to enhance management of delivery processes using artificial intelligence.
The online giant is introducing three tools built with advanced AI functionality designed to streamline and improve the accuracy of delivery operations. Following are brief overviews of each solution:
Wellspring
This generative AI solution is intended to improve delivery accuracy for customers. Wellspring harnesses data from dozens of sources, including satellite imagery, road networks, building footprints, customer instructions, information from prior deliveries, and street imagery, to help drivers better navigate complex and varied environments such as multi-building apartment complexes or brand-new neighborhoods that don't yet appear on navigation apps.
Wellspring is designed to aid Amazon drivers in better identifying which apartment numbers correspond to a specific building in an apartment complex, which parking spots and entrance points provide the most convenient route for package drop-off, and the location of shared mailrooms.
The technology also detects building entrances and mailroom locations by analyzing proof-of-delivery photos and location data from past deliveries.
When Amazon started testing Wellspring in the U.S. in October 2024, the system mapped over 2.8 million apartment addresses to their corresponding buildings across more than 14,000 complexes, while also identifying parking locations at 4 million addresses.
AI-based demand forecasting model
Amazon’s supply chain has a new foundational AI forecasting model designed to predict what customers will want, where they’ll want it, and when, for hundreds of millions of products per day.
Previous Amazon systems used sales history to guide inventory planning decisions, and its new AI foundation model adds time-bound data like weather patterns and holiday schedules to place the right products in the right locations more accurately.
By analyzing regional differences like sunscreen sales in Cape Cod, Mass. in the summer months or ski goggles in Boulder, Colo. during peak ski season, Amazon says it is able to more accurately and efficiently cater to the different needs of individual communities we serve.
According to Amazon, this solution has contributed to a 10% improvement in long-term national forecasts for deal event and a 20% improvement in regional forecasts for millions of popular items. These results have helped the retailer boost productivity and shrinking its network’s carbon footprint.
In addition, Amazon says packages arrive faster, sometimes same-day instead of within two days, delivery partners travel fewer miles, traffic is reduced, and carbon emissions are avoided. Operations networks in the U.S., Canada, Mexico and Brazil are already using this technology, with future expansion planned soon.
[READ MORE: How Amazon achieved its fastest ever delivery speeds in 2024]
Agentic AI-equipped robots
A new agentic AI team within the Amazon robotics division will focus on building an AI framework to add the ability for robots to hear, understand natural language, reason about it, and act autonomously. By using vision language models (VLM) and policies that drive robotic actions, instructions can be issued in plain speak.
Amazon intends agentic AI capabilities to transform systems like its Proteus autonomous mobile robot that moves customer orders into versatile assistants capable of moving heavy objects in tight spaces, while freeing up human employees to work on higher-level tasks.
By developing agentic AI capabilities for its robotic fleet, Amazon also hopes to create safer jobs for front-line employees who can let robots handle repetitive tasks; enable faster delivery for customers as robots are rerouted to where they’re needed most; and create better efficiencies as one robot can perform multiple jobs.
"These are just a few of the many ways we’re using AI to improve our customer, employee, and partner experience, as AI changes every aspect of how we work," Amazon said in a corporate blog post.
