Key takeaways

  • Oracle Fusion Claw helps Fusion Agentic Applications handle complex work, process large datasets, and adapt plans as conditions change.
  • Businesses set permissions and approval requirements, with controls enforced during execution and an auditable record of each outcome.
  • Release 26D of Oracle Fusion Cloud Applications introduces 25 Claw-powered applications across finance, HR, supply chain, and sales.

Every business faces complex problems that require specialist expertise. Consider a nurse administrator building a staffing schedule while balancing qualifications, availability, labor rules, and overtime. Or a supply chain planner who models constraints and alternatives across thousands of variables. Traditionally, specialists have done this work using a mix of enterprise applications, spreadsheets, and add-on tools.

Oracle Fusion Agentic Applications began changing how specialists work by coordinating their tasks to achieve a business outcome. Oracle Fusion Claw extends that capability to more complex, long-running work, such as creating a complicated staffing schedule or inventory plan. It gives Fusion Agentic Applications the ability to handle large data sets, revise plans as conditions change, and carry out authorized actions while the business defines the boundaries for what the application can do. This article explains how Fusion Claw works, the governance that supports it, and how it can be used in finance, HR, supply chain, and the front office.

How Fusion Claw handles complex work

Fusion Claw is a governed execution runtime that enables Fusion Agentic Applications to handle enterprise-scale work spanning many steps, large data sets, and changing conditions by combining AI reasoning with deterministic computation. The AI reasoning portion uses an advanced large language model (LLM) to understand the objective, analyze the problem, evaluate alternatives, and create a plan. Deterministic computation then does the repetitive part—calculations, reconciliations, and simulations across large sets of data. This two-part approach can lower costs because AI is used only for the parts of the work that require reasoning, while high performance computing handles large-scale, repetitive tasks.

Once it begins working toward an outcome, Fusion Claw can revise the plan when conditions change without requiring the user to remain involved at every step. There’s continuity across runs, too. Fusion Claw preserves context from completed work so similar tasks can build on earlier results.

More autonomy within boundaries set by the business

As Fusion Claw works more autonomously, an important question arises: What is it allowed to do? Oracle addresses that with an architecture separating the AI reasoning layer from the authority and execution layers. This prevents the LLM from giving itself additional permissions. A second boundary isolates the deterministic execution inside a secure, temporary environment, preventing the LLM from accessing—and interfering with—the underlying data and system of record.

Here’s a high-level overview of how the governance works:

  • Organizations set the objectives, policies, standard operating procedures, permissions, approval requirements, and escalation points in an Enterprise Operating Envelope. This is the authority layer that defines the boundaries for the work.
  • During execution (in the isolated environment), an Outcome Trust Harness enforces those controls. It governs which data and capabilities Fusion Claw can use, which actions it can take, and when approval or escalation is required.
  • When the work is complete, an Outcome Receipt provides an auditable record of the outcome. It shows the evidence used, decisions made, actions taken, and transactions affected.

Organizations also decide how much autonomy to grant each Fusion Agentic Application. They can require approval before actions are taken, allow Fusion Claw to work while people direct or intervene as needed, or permit fully autonomous execution within defined permissions and controls. In each case, the business—not the LLM—sets the limits.

Applying Fusion Claw across the business

Oracle is introducing 25 Claw-powered Fusion Agentic Applications with Release 26D. These examples show how they work across the business:

  • In finance, Ledger Workspace can improve close readiness by reconciling large volumes of accounting entries, investigating anomalies, and applying deterministic computation to complex accounting work.
  • In HR, Workforce Operations Command Center can create staffing plans that meet service requirements while balancing skills, availability, schedules, labor rules, and costs.
  • In supply chain, Shipment Load Planning Workspace can reduce shipping costs by comparing different consolidation options and executing the plan when authorized.
  • In sales, Account Territory Growth Plan can efficiently allocate opportunities by evaluating territory configurations and account coverage.

From recommendations to outcomes

Fusion Claw lets Fusion Agentic Applications take on more complex, long-running work to achieve business outcomes. People still define the objectives and controls, but more of the specialist work can be delegated to the application within those boundaries. For Fusion Apps customers, the opportunity is to look at recurring tasks that take significant effort to investigate, plan, and carry through. Then decide where Fusion Claw could take on more of that work while keeping people in control.

If you’d like to learn more, download the technical paper “Oracle Fusion Claw.”

Related posts you might like

Get more information link

If you’re an Oracle customer and want to get new stories from The Fusion Insider by email, sign up for Oracle Cloud Customer Connect. If you’re an Oracle partner and want to learn more, visit the Oracle Partner Community.