Agentic AI Heats up the Supply Chain Software Space—Concise Market Messaging Is Imperative to Gain Momentum

Agentic Artificial Intelligence (AI) has been a primary topic of discussion this year in the supply chain management space. In May, supply chain software giant Blue Yonder launched five agents as a part of its new Cognitive Solutions. With the intention to enable end users to act with machine speed and precision, these AI agents are built on more than 20 years of AI and Machine Learning (ML) research. The following are the five AI agents Blue Yonder launched:

• Inventory Ops Agent: This agent identifies mismatches and assesses root causes between supply and demand, whether from constraints or errors. With this information, it suggests fixes like alternate sourcing or demand swaps and communicates these to downstream partners.
• Shelf Ops Agent: This agent increases productivity across multiple stores and projects. It speeds up planogram editing—the visual layouts used for product placement in stores—by allowing planners to make updates through simple, natural language commands.
• Logistics Ops Agent: This agent monitors transport conditions in real time, automates scheduling, and suggests routing changes.
• Warehouse Ops Agent: This agent coordinates shifting warehouse activities to deliver consistency and responsiveness. It reallocates labor, adjusts layouts, detects outbound risks, and manages One Time in Full (OTIF) compliance to raise throughput and service quality.
• Network Ops Agent: This agent enables customers to oversee their multi-enterprise operations by providing predictive updates, carrier and order automation, and disruption management. It allows cross-channel interactions to jointly address problems and fine-tune performance.

Blue Yonder’s leadership highlighted that the company had rewritten 28 different planning applications onto one platform, which will be a huge enabler to deploy these agents.

Blue Yonder isn’t the only supply chain management solutions provider with Agentic AI innovations. Manhattan Associates also announced a suite of digital agents integrated into the Active platform, which sits within Manhattan’s cloud-native microservices architecture.  Its AI agents are Intelligent Store Manager, Labor Optimizer Agent, Wave Inventory Research Agent, Contextual Data Assistant, and Virtual Configuration Consultant. These agents autonomously manage tasks, adapt to changes, and optimize workflows using Large Language Models (LLMs). In addition, Manhattan also launched Agent Foundry, which is a platform within the ecosystem that allows its customers to create specialized agents tailored to their unique processes and preferences to address custom needs. Customers can also rely on Manhattan or third-party partners to build these specialized agents.

Software giant SAP also joined in on the action. At its Sapphire event, SAP announced that its AI copilot, Joule, will enable planners to swiftly interpret supply chain planning results through intelligent analysis of demand, inventory, and supply planning. In addition, it also announced a few key AI agents:

• Maintenance Planner Agent: Aims to streamline maintenance planning and coordination.
• Shopfloor Supervisor Agent: Proactively mitigates shop-floor disruptions, such as machine failures, and performs necessary rescheduling to maintain productivity.
• Field Service Dispatcher Agent: Autonomously schedules and optimizes service orders within SAP’s Field Service Management platform.

While the last couple of years have been filled with Generative Artificial Intelligence (Gen AI) innovations, AI agents and Agentic AI systems have been the biggest innovation this year, as showcased by multiple vendors. While end users now recognize the benefits of Gen AI, it is important to draw the distinction from an application and benefit perspective. When it comes to Gen AI, LLMs are the main driver and the chatbot is the interface that interacts with the user. It leverages models with large memories, and the content is generated based on what has been learned. In a supply chain management context, Gen AI is typically used for one-shot queries like report generation or information retrieval. It’s reactive and doesn’t retain context, which limits its application in dynamic real-time environments.

It is also important to distinguish between Agentic AI systems and AI agents. When it comes to Agentic AI systems, LLMs are a tool utilized to reason with users, interfere when needed, and generate insights. Meanwhile, AI agents leverage Retrieval-Augmented Generation (RAG) to draw data from operations in real time, which can enable business process improvements. AI agents are pro-active, goal-driven systems that can reason, plan, and act autonomously in real time. They operate dynamically and can engage in back-and-forth conversations with users. This means they can adapt in real time to changing conditions such as demand shifts, labor availability, and weather. Enterprises and solutions ecosystems are increasingly adopting Agentic AI systems. Enterprises can tackle complex issues involving multiple stakeholders like rerouting goods or resolving supplier delays. Essentially, this means less manual intervention and faster issue resolution, potentially fostering a culture of smarter decision-making.

With the entire industry talking about Agentic AI, there is a need for end users to streamline the deployment to ensure maximum Return on Investment (ROI). From an ROI perspective, users need to prioritize areas where human decision fatigue or delays are the costliest. They should focus on specific, high-value pain points that can provide quick wins. Demand forecast support, order exception handling, identifying data silos or whitespaces, and supplier coordination are a few use cases that can be low-hanging fruit. Data integrity and integration are also important aspects of implementation. AI agents are only as effective as the data they can be trained on. There is a need to standardize data being migrated from disparate digital solutions via robust data governance.

For vendors like Blue Yonder, SAP, and Manhattan Associates, the primary strategic focus should be to show their end-to-end capabilities. Their market messaging needs to stress how Agentic AI can enable “reactive to anticipatory” decision-making. The fact that AI agents can act within enterprise contexts and can be curated according to custom needs because of the data they are trained on should also be highlighted more. More importantly, it is important to note that we are still at the early stages of Agentic AI. As a recent ABI Insight highlights, positioning Agentic AI as modular plug-and-play enhancements, rather than complete overhauls that could lead to organizational inertia, could be key. As a result, displaying a clear decision rationale behind the AI models and showcasing how AI can enhance decision-making with the use of confidence scores can be an ideal way to build interest in the market. Agentic AI is showing a lot of promise in the supply chain management space, so highlighting the use cases that resonate with end users the most will be imperative to build traction.