As companies seek new use cases for robots in their supply chains, many will find themselves needing to integrate different robots from different vendors that perform different tasks.
The past few years have witnessed an explosion in interest in and use of robots within the supply chain. According to Gartner’s most recent “Supply Chain Technology User Wants and Needs Study”—a cross-industry, cross-company size, and cross-geography survey—a staggering 96% of respondents said they were investing, or were planning to invest, in robotics over the next two years.
Of those respondents, 7% had already fully deployed a robotics solution, and 29% were currently deploying one. Furthermore 93% of these current robot users said that they planned to expand the fleet size of their existing robot platforms, and 94% said they were pursuing additional use cases for robotics in their operations.
As a result of this interest, we believe there will be exponential growth in what we call the “intralogistics smart robots” (ISR) marketplace over the next decade. In fact, Gartner predicts that by 2028, 50% of large enterprises will have adopted some form of ISRs in their warehouse or manufacturing operations.
Currently, Gartner tracks 34 different categories of intralogistics smart robots. Among the most relevant categories to logistics leaders are six that we see companies deploying and having the most success with today. The categories include:
Basic transport—This category involves autonomous mobile robots (AMRs) designed to move goods around warehouses and plants. These trainable and intelligent robots excel at moving goods long distances and can operate through multiple shifts. By using them, companies reduce the amount of time human workers waste traveling across the warehouse.
Collaborative picking (robot to person)—For this application, a human worker and robot move through the warehouse together. The human worker picks products into bins or totes transported by the robot. This category will enhance human labor by improving picker efficiency, cycle time, and throughput. This “cobot” augmentation of human effort can leverage existing infrastructure and will be utilized in high volume/velocity picking environments such as e-commerce.
Goods to person (G2P)—For this variation, the robots deliver multiple goods on mobile storage units (MSUs) to a stationary area, where a human picks goods for multiple orders onto another MSU. When all the orders are complete, robots deliver the MSUs to packing stations. G2P robots can eliminate wasted travel time for human workers, reducing drudgery and fatigue.
Sortation robots—This category will be seen both in e-commerce and parcel-sorting environments and will improve order-fulfillment accuracy and agility while streamlining picking and packing operations. These robots can replace powered conveyors and are not bolted to the floor, meaning they are adaptable and reconfigurable on demand, with lower fixed infrastructure requirements.
Robotic picking—These robotic solutions are designed to handle the most mundane pick-and-place tasks. These solutions combine robotic arms, different forms of end effectors or grippers, and 3D vision systems, all enabled by advanced machine learning and artificial intelligence (AI). Robotic picking works particularly well in environments where the items are a consistent size and shape.
Cube robotic G2P systems—In this category, G2P robots autonomously move goods in totes or cases that are stored in a pre-built cube/grid structure. The robots also deliver these totes or cases at the appropriate time to humans at pick stations. These solutions work well for high-density and high-speed environments. For example, these applications work well in facilities with large quantities of small items that are ordered frequently. This category is scalable and adaptable and is typically delivered as a larger integrated system.
Each of these categories represents different use cases and operating models, some are designed to be stationary, others mobile; some are designed to operate alone and autonomously; while others are designed to complement human labor.
As companies seek new use cases, many will start to have different robots from different vendors performing different tasks. We believe that within the next five years more than 40% of large enterprises will have a heterogeneous fleet of ISRs in their warehouse operations. The good news is that many companies will begin to leverage ISRs in their operations. The bad news is that this creates challenges for companies. Namely, how do they integrate with and orchestrate the work of a heterogeneous fleet of robots? And how do they coordinate between different fleets?
Standardized software needed
To integrate and orchestrate the work of a fleet of heterogenous robots, companies will need standardized software that can easily unite a variety of agents and robot platforms. Gartner refers to this emerging software as “multiagent orchestration platforms.” These solutions act like intelligent middleware that integrate and orchestrate work among various business applications, heterogenous fleets of operational robots, and other automated agents like doors or elevators. These solutions will assign work to the right robots based on the characteristics of the immediate tasks and will orchestrate communication between different robot platforms and other types of automation agents. (See Figure 1.)
This type of software becomes increasingly necessary as the robotic environment becomes more complex. When companies invest in their first ISR platform, they will typically just create a one-off connection between their business applications—such as a warehouse management system (WMS)—and their robot provider’s fleet management system. This, while not optimal, works for one robot. However, as a company’s fleet of robots grows, simple point-to-point API (application programing interface) integration will not be enough. Companies will need an orchestration capability that can assign work to the right robots based on near-real-time information. These work assignments will need to take into consideration the characteristics of the activity and the capabilities of various automation agents. A multiagent orchestration platform will reduce the time, effort, and cost to onboard new robots. It will also reduce support cost, ultimately making organizations more efficient because work will be assigned to the robot best suited for the task. As a result of this need, we believe that by 2026, more than 50% of companies deploying intralogistics robots will adopt a multiagent orchestration platform.
Of course, most companies will not recognize the need for these types of solutions until they move beyond one or two robot platforms. Then, they may attempt to find a solution through their current WMS provider or their robot provider’s fleet management systems. These systems may or may not address the need for orchestration and integration across and between a variety of robot platforms. While some providers do offer these types of orchestration platforms, many ISR providers’ fleet management solutions are largely focused in and around their own robot offerings and are not true multiagent orchestration platforms. To be sure, many ISR providers are focusing more on software, as they are concerned that they will be commoditized by less expensive robot hardware. But we do not expect a universal fleet management platform that works across robot platforms any time soon, if ever. Consequently, for the foreseeable future, companies with heterogeneous fleets of robots will need a multiagent orchestration platform.
To identify the right platform capable of orchestrating and integrating their heterogenous robot fleet, companies should start by analyzing the integration requirements as their robot fleet expands beyond a single vendor. Along with that, they should study how work will be assigned to the various robots and other automation agents and determine what orchestration logic will be needed to support this simultaneously. Once they understand their orchestration and integration requirements, they should then look for the multiagent orchestration platform that best addresses their needs.
The launch is based on “Amazon Nova,” the company’s new generation of foundation models, the company said in a blog post. Data scientists use foundation models (FMs) to develop machine learning (ML) platforms more quickly than starting from scratch, allowing them to create artificial intelligence applications capable of performing a wide variety of general tasks, since they were trained on a broad spectrum of generalized data, Amazon says.
The new models are integrated with Amazon Bedrock, a managed service that makes FMs from AI companies and Amazon available for use through a single API. Using Amazon Bedrock, customers can experiment with and evaluate Amazon Nova models, as well as other FMs, to determine the best model for an application.
Calling the launch “the next step in our AI journey,” the company says Amazon Nova has the ability to process text, image, and video as prompts, so customers can use Amazon Nova-powered generative AI applications to understand videos, charts, and documents, or to generate videos and other multimedia content.
“Inside Amazon, we have about 1,000 Gen AI applications in motion, and we’ve had a bird’s-eye view of what application builders are still grappling with,” Rohit Prasad, SVP of Amazon Artificial General Intelligence, said in a release. “Our new Amazon Nova models are intended to help with these challenges for internal and external builders, and provide compelling intelligence and content generation while also delivering meaningful progress on latency, cost-effectiveness, customization, information grounding, and agentic capabilities.”
The new Amazon Nova models available in Amazon Bedrock include:
Amazon Nova Micro, a text-only model that delivers the lowest latency responses at very low cost.
Amazon Nova Lite, a very low-cost multimodal model that is lightning fast for processing image, video, and text inputs.
Amazon Nova Pro, a highly capable multimodal model with the best combination of accuracy, speed, and cost for a wide range of tasks.
Amazon Nova Premier, the most capable of Amazon’s multimodal models for complex reasoning tasks and for use as the best teacher for distilling custom models
Amazon Nova Canvas, a state-of-the-art image generation model.
Amazon Nova Reel, a state-of-the-art video generation model that can transform a single image input into a brief video with the prompt: dolly forward.
Economic activity in the logistics industry expanded in November, continuing a steady growth pattern that began earlier this year and signaling a return to seasonality after several years of fluctuating conditions, according to the latest Logistics Managers’ Index report (LMI), released today.
The November LMI registered 58.4, down slightly from October’s reading of 58.9, which was the highest level in two years. The LMI is a monthly gauge of business conditions across warehousing and logistics markets; a reading above 50 indicates growth and a reading below 50 indicates contraction.
“The overall index has been very consistent in the past three months, with readings of 58.6, 58.9, and 58.4,” LMI analyst Zac Rogers, associate professor of supply chain management at Colorado State University, wrote in the November LMI report. “This plateau is slightly higher than a similar plateau of consistency earlier in the year when May to August saw four readings between 55.3 and 56.4. Seasonally speaking, it is consistent that this later year run of readings would be the highest all year.”
Separately, Rogers said the end-of-year growth reflects the return to a healthy holiday peak, which started when inventory levels expanded in late summer and early fall as retailers began stocking up to meet consumer demand. Pandemic-driven shifts in consumer buying behavior, inflation, and economic uncertainty contributed to volatile peak season conditions over the past four years, with the LMI swinging from record-high growth in late 2020 and 2021 to slower growth in 2022 and contraction in 2023.
“The LMI contracted at this time a year ago, so basically [there was] no peak season,” Rogers said, citing inflation as a drag on demand. “To have a normal November … [really] for the first time in five years, justifies what we’ve seen all these companies doing—building up inventory in a sustainable, seasonal way.
“Based on what we’re seeing, a lot of supply chains called it right and were ready for healthy holiday season, so far.”
The LMI has remained in the mid to high 50s range since January—with the exception of April, when the index dipped to 52.9—signaling strong and consistent demand for warehousing and transportation services.
The LMI is a monthly survey of logistics managers from across the country. It tracks industry growth overall and across eight areas: inventory levels and costs; warehousing capacity, utilization, and prices; and transportation capacity, utilization, and prices. The report is released monthly by researchers from Arizona State University, Colorado State University, Rochester Institute of Technology, Rutgers University, and the University of Nevada, Reno, in conjunction with the Council of Supply Chain Management Professionals (CSCMP).
Specifically, 48% of respondents identified rising tariffs and trade barriers as their top concern, followed by supply chain disruptions at 45% and geopolitical instability at 41%. Moreover, tariffs and trade barriers ranked as the priority issue regardless of company size, as respondents at companies with less than 250 employees, 251-500, 501-1,000, 1,001-50,000 and 50,000+ employees all cited it as the most significant issue they are currently facing.
“Evolving tariffs and trade policies are one of a number of complex issues requiring organizations to build more resilience into their supply chains through compliance, technology and strategic planning,” Jackson Wood, Director, Industry Strategy at Descartes, said in a release. “With the potential for the incoming U.S. administration to impose new and additional tariffs on a wide variety of goods and countries of origin, U.S. importers may need to significantly re-engineer their sourcing strategies to mitigate potentially higher costs.”
Grocers and retailers are struggling to get their systems back online just before the winter holiday peak, following a software hack that hit the supply chain software provider Blue Yonder this week.
The ransomware attack is snarling inventory distribution patterns because of its impact on systems such as the employee scheduling system for coffee stalwart Starbucks, according to a published report. Scottsdale, Arizona-based Blue Yonder provides a wide range of supply chain software, including warehouse management system (WMS), transportation management system (TMS), order management and commerce, network and control tower, returns management, and others.
Blue Yonder today acknowledged the disruptions, saying they were the result of a ransomware incident affecting its managed services hosted environment. The company has established a dedicated cybersecurity incident update webpage to communicate its recovery progress, but it had not been updated for nearly two days as of Tuesday afternoon. “Since learning of the incident, the Blue Yonder team has been working diligently together with external cybersecurity firms to make progress in their recovery process. We have implemented several defensive and forensic protocols,” a Blue Yonder spokesperson said in an email.
The timing of the attack suggests that hackers may have targeted Blue Yonder in a calculated attack based on the upcoming Thanksgiving break, since many U.S. organizations downsize their security staffing on holidays and weekends, according to a statement from Dan Lattimer, VP of Semperis, a New Jersey-based computer and network security firm.
“While details on the specifics of the Blue Yonder attack are scant, it is yet another reminder how damaging supply chain disruptions become when suppliers are taken offline. Kudos to Blue Yonder for dealing with this cyberattack head on but we still don’t know how far reaching the business disruptions will be in the UK, U.S. and other countries,” Lattimer said. “Now is time for organizations to fight back against threat actors. Deciding whether or not to pay a ransom is a personal decision that each company has to make, but paying emboldens threat actors and throws more fuel onto an already burning inferno. Simply, it doesn’t pay-to-pay,” he said.
The incident closely followed an unrelated cybersecurity issue at the grocery giant Ahold Delhaize, which has been recovering from impacts to the Stop & Shop chain that it across the U.S. Northeast region. In a statement apologizing to customers for the inconvenience of the cybersecurity issue, Netherlands-based Ahold Delhaize said its top priority is the security of its customers, associates and partners, and that the company’s internal IT security staff was working with external cybersecurity experts and law enforcement to speed recovery. “Our teams are taking steps to assess and mitigate the issue. This includes taking some systems offline to help protect them. This issue and subsequent mitigating actions have affected certain Ahold Delhaize USA brands and services including a number of pharmacies and certain e-commerce operations,” the company said.
Editor's note:This article was revised on November 27 to indicate that the cybersecurity issue at Ahold Delhaize was unrelated to the Blue Yonder hack.
The new funding brings Amazon's total investment in Anthropic to $8 billion, while maintaining the e-commerce giant’s position as a minority investor, according to Anthropic. The partnership was launched in 2023, when Amazon invested its first $4 billion round in the firm.
Anthropic’s “Claude” family of AI assistant models is available on AWS’s Amazon Bedrock, which is a cloud-based managed service that lets companies build specialized generative AI applications by choosing from an array of foundation models (FMs) developed by AI providers like AI21 Labs, Anthropic, Cohere, Meta, Mistral AI, Stability AI, and Amazon itself.
According to Amazon, tens of thousands of customers, from startups to enterprises and government institutions, are currently running their generative AI workloads using Anthropic’s models in the AWS cloud. Those GenAI tools are powering tasks such as customer service chatbots, coding assistants, translation applications, drug discovery, engineering design, and complex business processes.
"The response from AWS customers who are developing generative AI applications powered by Anthropic in Amazon Bedrock has been remarkable," Matt Garman, AWS CEO, said in a release. "By continuing to deploy Anthropic models in Amazon Bedrock and collaborating with Anthropic on the development of our custom Trainium chips, we’ll keep pushing the boundaries of what customers can achieve with generative AI technologies. We’ve been impressed by Anthropic’s pace of innovation and commitment to responsible development of generative AI, and look forward to deepening our collaboration."