Recent research on AI Agents in the supply chain field has been conducted using the latest Agent features of ChatGPT-5 for search studies. ChatGPT operates with remarkable efficiency, thoroughly searching through the latest reports and articles from major consulting firms, achieving high accuracy and in-depth reading satisfaction. However, the insights from the latest reports of various consulting firms are not particularly novel, and the summary below is provided for reference. Some content from consulting firms like Bain and Roland Berger was omitted due to its lack of substance.
（As external links are not allowed in public accounts, please search for detailed reports by their titles.）
McKinsey
“How Gen AI is Reshaping Supply Chains” – April 2025
Key Insight: “McKinsey points out that generative AI can not only generate text but also assist humans in real-time within the supply chain through agents like ‘virtual schedulers’. For example, the virtual scheduler acts as a ‘driver assistant’, interacting with drivers and dispatchers to reduce logistics delays and save costs; another AI engine can converse with order managers to automatically extract and optimize order allocation strategies using historical rules, making algorithms part of the team, requiring human planners for training. When implementing generative AI, companies should identify high-potential use cases, build flexible data and technology foundations, and consider changes in employee work methods.“
“AI in the Workplace: A Report for 2025” – January 2025
Key Insight: “The report summarizes ‘AI Agents’, suggesting that by 2025, agents will no longer be limited to chatbots but will be capable of completing multi-step tasks, such as processing payments and shipments directly after conversing with customers; Salesforce’s Agentforce can deploy autonomous agents that can complete complex workflows.”
“Seizing the Agentic AI Advantage” – June 2025
Key Insight: “McKinsey emphasizes the need to break through the ‘generative AI paradox’ by shifting AI from chat modes to vertical scenarios, enabling collaboration through AI Agents across the end-to-end supply chain. Agents serve as the ‘command layer’ of the supply chain, connecting internal planning systems with external data such as weather and demand, predicting demand, identifying delays, replanning transportation and inventory, and selecting optimal transportation modes, thereby improving service levels while reducing costs and carbon emissions. Deploying agents requires redesigning processes rather than simply embedding them.”
BCG
“AI Agents: What They Are and Their Business Impact” – Early 2025
Key Insight: “BCG describes AI Agents as systems with goal orientation, multi-step planning, autonomous execution, and adaptability, noting that most ‘agents’ are still in prototype stages. Barriers include: models being difficult to generalize, decision processes being hard to explain, needing training through real workflows, collaboration issues among multiple agents, and lack of data/tools. It is recommended that companies prioritize experimenting with agents in high-return scenarios, quickly iterate, prepare data and connection interfaces, and cultivate a learning organization.”
“Frontier Technologies in Industrial Operations: The Rise of Artificial Intelligence Agents” – February 2025
Key Insight: “The white paper distinguishes between two types of agent-based AI: virtual agents (software that autonomously achieves goals and directs factory processes in a digital environment) and embodied agents (robots with perception and action capabilities). It argues that agent-based AI can address labor shortages, sustainability, and high customization demands. Manufacturing companies should adopt a value-driven approach, achieve IT/OT integration, build 5G and other infrastructures, and prepare organizations by enhancing employee skills and establishing an AI culture.”
“AI at Work 2025: Momentum Builds, but Gaps Remain” – Early 2025
Key Insight: “Research shows that only about 13% of employees believe AI Agents are deeply integrated into workflows, but most hold a positive attitude after understanding agent capabilities. BCG recommends companies pilot agents and track value, invest in training, and change workflows to unlock potential.”
Accenture
“Making Autonomous Supply Chains Real” – First Half of 2025
Key Insight: “Accenture believes that future autonomous supply chains will integrate automation (machines replacing humans) and delegation (machines making independent decisions), with human-machine collaboration remaining key. Advanced AI Agents will require end-to-end visibility, with transparent enterprise data as the basis for decision-making. Research shows that 66% of surveyed companies plan to advance supply chain autonomy by 2035, with 40% aiming for systems to handle most operational decisions independently.” To build autonomous supply chains, Accenture proposes three actions:1. Build a unified and secure data core that aggregates inventory, sales, and forecasting data in real-time, allowing AI to optimize decisions promptly during disruptions;2. Invest in appropriate AI technologies and redesign processes, upgrade legacy systems, and build a flexible AI technology stack. Agents can automatically handle routine tasks, provide real-time insights, predict disruptions, and suggest solutions, helping to break down information silos and improve efficiency;3. Restructure human-machine collaboration models, enabling multifunctional teams to collaborate faster through a platform-based operational model and reshape personnel structures.”
“Accenture Expands AI Refinery and Launches New Industry Agent Solutions” – March 2025
Key Insight: “Accenture announced the addition of an ‘Agent Builder’ to its AI Refinery platform, allowing business personnel to quickly create and customize AI agents without coding, with plans to launch over 100 industry-specific agent solutions within the year. These agents utilize NVIDIA’s inference models to automate processes and improve efficiency across industries such as telecommunications, financial services, insurance, manufacturing, and healthcare. Examples include:• Call Center Agents: Track conversations, provide real-time suggestions, and help agents handle calls more efficiently, expected to increase call handling speed by 25 times;• Insurance Underwriting Agents: Automatically read, analyze application information, and assess risks, increasing underwriting speed and ensuring 100% of applications are processed;• Financial ‘Order to Cash’ Agents: Automatically handle order validation, invoice reconciliation, and receivables management, allowing finance personnel to focus more on strategic decision-making. Accenture’s research indicates that about one-third of organizations have scaled at least one industry-specific solution in core processes, increasing the likelihood of achieving ROI beyond expectations by three times.”
Deloitte
“Three New AI Breakthroughs Shaping 2026” – 2025
Key Insight: “Deloitte defines ‘agentic AI’ as autonomous systems that can adapt to environments, make complex decisions, and collaborate with humans or other agents, capable of automatically handling dynamic, multi-step processes. The article notes that AI Agents have prospects in customer service, supply chain (real-time optimization of inventory, logistics, and procurement), and finance. Research shows that most organizations are still in pilot or undeployed stages, with a few large tech companies having moderate deployments; a LinkedIn survey indicates that nearly half of respondents believe agents will significantly change organizations within 2-3 years. Predictions for 2026 include: ① agents will expand from pilot to production applications, especially with the emergence of ‘out-of-the-box’ solutions; ② governance and compliance will become a focus, requiring companies to establish clear frameworks; ③ companies will invest in cultivating new roles such as ‘agent ops’ to provide employees with skills to work alongside agents.”
“Deloitte Unveils Zora AI” – May 2025
Key Insight: “Deloitte launched the ‘Zora AI’ platform for building perceptive, reasoning, and action-capable digital agents, equipped with functional and industry-specific knowledge. The platform is based on NVIDIA’s Llama Nemotron model and NVIDIA AI stack, allowing for rapid deployment and integration with existing systems. Zora AI includes agents for finance, human capital, supply chain, procurement, sales and marketing, and customer service, working collaboratively with employees as a multi-agent system. Deloitte plans to use Zora AI internally to reduce expense management costs by 25%, improve productivity by 40%, and offer it as a cloud subscription model. The platform adheres to trusted AI principles and incorporates human feedback loops to enhance transparency and explainability.”
IBM
“AI Projects to Profits” – June 2025
Key Insight: “Research interviewed 2,900 executives, finding that companies expect AI-driven workflows to increase from 3% to 25% by the end of 2025, many driven by AI Agents. 70% of respondents believe AI Agents are crucial for the future of enterprises, and 83% expect AI Agents to enhance process efficiency and output before 2026, with 71% believing agents will be able to autonomously adapt to changing workflows. Executives noted that the main benefits of agent-based AI include: improved decision-making (69%), cost reduction through automation (67%), gaining competitive advantage (47%), enhancing employee experience (44%), and increasing talent retention (42%). However, they also expressed concerns that data quality, trust, and talent gaps remain barriers to adoption. IBM’s leaders emphasized that deploying agents is not merely about embedding them into existing processes but requires redesigning processes, reconstructing user experiences, and orchestrating agents end-to-end.”
In summary, the consensus among consulting firms is that AI Agents are gradually moving from concept to implementation, with the supply chain, logistics, and manufacturing sectors being the most promising application scenarios due to their data-intensive, complex processes and high labor costs. Companies need to prepare in terms of data infrastructure, process redesign, and talent development, gradually expanding after validating value through pilot projects to achieve autonomous supply chains and intelligent manufacturing through human-machine collaboration.
Related Original Long Reads:AI Agents Will Not Automatically Create Order: Why Process Foundations Determine AI SuccessThe Talent Model for Supply Chains in the AI Era: Becoming the Da Vinci of the ‘New Renaissance’The ‘Navigational Chart’ for Landing on New Continents: Gartner’s Supply Chain AI Case LibraryIn these turbulent times, reading long articles is not easy; you are unique!If this article has been helpful to you, or if you are interested in the supply chain and logistics field, don’t forget tolikeit, share it in your circle of friendsto more friends. Every like and share is the greatest support for the author and helps more people understand the dynamics and trends of the industry.