TL;DR — Key Takeaways
- Amazon is exploring Project Tetromino, an AI- and robotics-driven concept designed to automate more sorting and staging work at delivery stations.
- Internal planning documents reportedly suggest Tetromino could process packages 2.5 times faster than current delivery stations, although Amazon disputes the reported investment figures and rollout timeline.
- One technology being considered comes from Boxbot, whose system uses high-density automated storage and sequencing to organize packages for vehicle loading.
Amazon is examining how far it can automate the final stop in its delivery network with a concept that could process packages 2.5 times faster than current delivery stations.
Project Tetromino, as it is internally codenamed, would use AI and robotics to automate more of the sorting and staging that takes place before packages are loaded for delivery. The concept was detailed in a leaked internal Amazon planning document, as reported by Business Insider.
The document suggests Amazon could invest $103 million in a Tetromino pilot site in 2028, followed by five additional sites in 2029 and 10 more in 2030. This could bring its total investment in the project to over $530 million by 2029, according to the report. Amazon disputed those figures and timelines, however. A spokesperson told Business Insider that Tetromino is still an early-stage concept and said the document’s financial projections and rollout schedule are inaccurate and do not reflect the company’s current plans.
But even if the details change, the document provides a rare look at how Amazon is thinking about automating its delivery stations, the last stop in its operations network before packages head out to customers. These facilities run about 40 operational processes, including receiving packages, sorting them into final delivery routes, packing them into bags and staging them for pickup by delivery service partners. Because packages must be organized around specific departure windows and vehicle loads, staging is a time-sensitive step in the delivery process that seems primed for further automation.
One technology under consideration for Tetromino is a system from supply chain robotics startup Boxbot, according to the report. Boxbot describes its approach as a form of automated Tetris, moving packages from conveyors into high-density storage, dynamically assigning them to available space and retrieving them in the sequence needed for loading based on delivery order and route requirements. The company specifically markets the system for last-mile delivery stations, where it says the technology can streamline vehicle loading and route dispatch.
Amazon has already been working to automate more of its delivery-station workflow. At its Last Mile Innovation Center in Dortmund, Germany, the company has tested what it calls its “delivery station of the future” using technologies including automated unloading equipment, conveyor-management software, computer vision capabilities and robotic sorting systems that route packages and bring them to workers for final handling. One projection-based system shows employees which packages belong in which route bags, showing where employees are still needed. Tetromino would go a step further by reducing those remaining manual handoffs.
How Tetromino might change staffing at delivery stations is still unclear. Amazon said in 2025 that each new rural U.S. delivery station creates an average of about 170 on-site jobs, not including additional driving opportunities through delivery service partners and Amazon Flex. The company has described its delivery-station automation endeavors as a way to make physically demanding work safer and easier while shifting employees to other tasks. Tetromino suggests the company is also examining how much of that work, and ultimately how much human labor, can be removed from the process altogether.
