Oracle adds AI task execution to its 26D supply-chain software release

Oracle’s latest supply-chain release puts AI closer to carrying out decisions, not just suggesting them. In its November 6 overview of Fusion Cloud SCM 26D, Oracle introduced Fusion Claw for analyzing problems, planning responses and completing work. The company says people retain control of goals, policies and important approvals.

From a readiness question to follow-up work

Oracle illustrates Fusion Claw with a product team asking whether a bill of material—the list of components needed for a product—is ready for launch. The system gathers relevant information, weighs options against business constraints and recommends next steps. It can then take permitted actions and monitor progress as conditions change.

That example describes more than a one-time answer: Oracle says the capability can carry follow-up work through to completion. In 26D, Fusion Claw powers capabilities in Product Readiness Workspace, Sourcing Command Center and Supply Planning Allocation Workspace. Those applications put its execution role inside product, purchasing and planning work.

The stated boundary is the business application’s control system. Oracle says Fusion Applications enforce permissions and business controls and keep an auditable record of actions. Its promise of less manual coordination therefore comes with an explicit division of responsibility: AI carries out authorized work, while people set goals, policies and key approvals.

Planning and procurement get more focused tools

Alongside Fusion Claw’s multistep work, the release adds tools aimed at specific operational tasks. Supply Chain Planning introduces centralized production campaign management to help teams sequence work and coordinate manufacturing schedules. Shelf-life planning matches supply to customers’ freshness requirements, while Business Growth Planning Workspace uses comparable launches to forecast new products with little historical data.

Procurement changes address communication and document handling. Users can start Microsoft Teams conversations directly from requisitions, purchase orders or intake requests, keeping discussions attached to the relevant work. Contract digitization extracts key information from existing documents to create managed contract records, rather than requiring teams to enter all that information manually.

Product data gets a separate set of changes. Product Lifecycle Management supports bulk item onboarding and updates through batches in Oracle’s Redwood interface. Teams can also inspect trading-partner item and manufacturer details within relationship views, which Oracle positions as a way to validate information with less navigation.

Controls reach the warehouse and factory floor

• Inventory teams can track stock by grade and expiration date. AI agents recommend packing configurations and alternate storage locations to reduce handling and manual lookups.
• Manufacturing adds approval workflows for changes to work definitions and barcode scanning for work-in-process inspections. Maintenance supports multiple labor assignments per technician.
• The Warehouse Execution Command Center highlights fulfillment priorities. Oracle says it helps teams address bottlenecks and use labor and equipment more effectively.

Other AI assistance targets exceptions and pricing. Teams can use it to diagnose and reprocess failed sales-order imports. The Price Administration Assistant supports promotions and discounts, checking request details with users before creating promotions. Here, the user check is built into the described task rather than left outside the workflow.

For investigations, Product Genealogy Workspace connects information across operations and suppliers to identify affected products. Quality Assurance Workspace is designed to detect emerging inspection risks and process instability. Oracle positions these applications around earlier intervention: connecting product relationships and spotting warning signs so teams can investigate and take corrective action.