Inside Oracle’s vision to redefine planning, manufacturing and global trade with Agentic AI

With these AI innovations, how is Oracle redefining the concept of a minimum viable supply chain, and why is this transition important for legacy enterprises?

Our vision is to enable what we describe as the autonomous supply chain. By autonomous, we mean supply chains that are self-aware, self-optimising, and self-executing. While this may sound aspirational, we are rapidly approaching a point where this level of automation is achievable.

This is important for several reasons, one of the most critical being disruption management. Disruptions can be small or large, positive or negative. Negative disruptions are often discussed, but positive disruptions are equally important. For example, when a new product performs significantly better in the market than expected, the supply chain must rapidly respond by increasing production and fulfilment.

Traditionally, managing these disruptions requires extensive manual effort. AI agents and agentic workflows can deliver dramatic productivity gains by automating large parts of this work.

A common example involves supplier component changes. A supplier may notify a manufacturer that a component is going end-of-life and will be replaced with a new version. Without AI, organisations must manually analyse existing inventory, work-in-process, affected products, and required engineering or manufacturing changes.

We developed an AI agent that begins by interpreting a PDF notification from the supplier. It identifies the impacted component, analyses all products that use it, assesses inventory levels across the supply chain, and evaluates work-in-process on the shop floor. What can take weeks for a customer to analyse manually can now be completed by an agent in minutes.

This reduction in response time significantly improves supply chain agility and resilience.

Can you explain the technical architecture behind these embedded AI capabilities and how they change daily workflows for supply chain professionals, making decision-making more rational and predictive?

AI agents manifest in several ways, but the most common is through a chat-based user experience where agents act as advisors.

Consider a customer return scenario. A customer service representative must understand company return policies, customer-specific contract terms, and eligibility conditions. The AI agent has access to this information and can provide recommendations on how to handle the return.

If the organisation is comfortable with the agent’s reliability, the agent can also execute actions autonomously. It can authorise the return, generate return shipping labels, and initiate replacement orders if required. These actions can occur entirely in the background, or users can interact with the agent directly to ask questions and instruct it to take specific actions.

This approach allows AI to augment human decision-making or fully automate processes where appropriate.

Many organisations start AI adoption with a pilot in one warehouse before scaling globally. What are the common breaking points when scaling AI, and how does Oracle help customers overcome them?

Many AI initiatives fail to progress beyond the pilot or proof-of-concept stage. We address this challenge in several ways.

Within our AI Agent Studio, we provide extensive evaluation and testing capabilities. Before an agent is deployed, whether in a pilot or at scale, we validate its behaviour through rigorous testing to ensure it produces the expected outcomes.

Once deployed, monitoring continues. We track response quality, performance, and user interactions. For example, if users consistently ignore or override an agent’s recommendations, that behaviour is captured. This allows us to investigate whether the issue lies in data quality, logic, or contextual understanding.

Enterprise supply chain applications require a high level of reliability and trust. Many AI pilots lack the rigour required for mission-critical environments. We provide foundational capabilities around validation, testing, security, and performance monitoring to ensure agents operate reliably both before and after deployment.

Do you see retailers increasingly using supply chain platforms as proving grounds for predictive analytics?

This trend is not limited to retail; two areas have seen the highest adoption of AI agents among customers.

The first is procurement, where AI supports sourcing and procuring goods and managing supplier interactions. The second is supply chain planning and demand planning.

We use advanced optimisation and forecasting techniques to generate recommendations, such as shipment strategies and demand forecasts. However, users may not always understand why a particular recommendation was produced.

To address this, we developed AI agents that analyse complex diagnostic data and explain the rationale behind optimisation results. These agents also identify data quality issues or parameter settings that may need adjustment.

This transparency helps planners build confidence in the system and continuously improve forecast and optimisation accuracy over time.

Supply chains are increasingly impacted by geopolitical instability and shifting trade alliances. How do AI-driven supply chain tools help organisations adapt and reroute operations under such conditions?

A critical requirement in global trade is accurate product classification. Classification codes determine tariffs, quota restrictions, and licensing requirements, and they vary across countries.

We developed an AI agent that assists with product classification by consuming regulatory content, which may change due to new trade agreements or geopolitical developments. The agent can reclassify products and assess the financial and operational impact of regulatory changes.

For example, it can identify when tariffs change from zero to 20 percent or vice versa. Global trade compliance is a complex discipline that traditionally relies on highly experienced professionals. AI agents can significantly accelerate this process while maintaining accuracy.

Manufacturing is increasingly moving closer to points of consumption. How does the integration of IoT and edge AI within Oracle Cloud support highly automated, localised smart factories?

We provide automated connectivity between our supply chain applications and shop-floor equipment. Our manufacturing modules are designed for workers on the factory floor, but they often require integration with additional systems that connect directly to equipment.

We have built partnerships with companies such as Litmus and Microsoft to enable IoT connectivity. This allows us to automate equipment configuration based on manufacturing work definitions, including machine settings required for specific products.

As equipment produces goods, quality measurements can be captured automatically to determine pass or fail outcomes. If a product fails inspection, the system can decide whether it should be scrapped or reworked.

This approach also supports configure-to-order manufacturing. Products can be centrally defined while allowing local variations based on market requirements. Configuration details are automatically communicated to individual machines, ensuring consistent execution across global manufacturing sites without manual intervention.

From a go-to-market perspective, what is Oracle’s strategy for the two announcements made today, and what are your expectations?

These were not brand-new products but enhancements to existing offerings. The first announcement focused on new AI agents across procurement, manufacturing, planning, and logistics. These enhancements align with our go-to-market strategy for key manufacturing industries, including industrial, automotive, life sciences, and medical devices.

The second announcement addressed enhancements across manufacturing, planning, order management, and inventory to support process manufacturing industries such as food and beverage, pharmaceuticals, chemicals, paints, steel, and related sectors.

These enhancements are designed to drive broader adoption within these vertical industries by addressing their specific operational complexities.