Oracle Fusion Finance: 12 New Autonomous Agents
- Michael Hulbert
- Apr 7
- 3 min read
Title: Oracle ERP
Date: 7 April 2026
Type: Blog
Author: Michael Hulbert (michael@saasiq.ai)
Word count: 1045 words
Reading time: 5 min
Published: 7 April 2026
Oracle announced 12 new Fusion Agentic Applications for finance and supply chain on 9 April 2026. These agents process invoices, collect outstanding payments, manage procurement auctions, and execute payroll changes autonomously. What separates them from copilots is native integration into Fusion and embedded security controls.
12 Agents, Three Finance Functions
Oracle shipped 12 pre-built agents that handle specific finance and supply chain workflows. The Accounts Payable agents process matching, three-way matching, and anomaly detection. The Claims Settlement Workspace handles payroll claim processing and recovery. The Collectors Workspace automates dunning and collection activities for outstanding payments. Procurement agents run sourcing auctions, evaluate bids, and recommend approval actions based on contract terms.
These agents are part of Oracle's broader 20+ agentic applications portfolio spanning ERP, HCM, Supply Chain, and customer experience modules. They execute work that finance teams currently handle with full-time headcount. Rather than assisting human users, these agents make transactional decisions, execute the transaction, and log the action in the audit trail.
Agentic Applications vs Copilots: The Critical Distinction
Every vendor is shipping copilots that draft invoices, suggest matching, and present options for human approval. Copilots reduce the time humans spend on routine work. Agentic applications perform the work without human intervention. The agents operate within defined guardrails: cost centre budgets, approval thresholds, supplier contracts, and regulatory constraints.
When an Accounts Payable agent processes an invoice, it executes the posting to the general ledger. Human approval is no longer in the workflow.
This is a different value proposition. Copilots save individuals time on their allocated work. Agentic applications reduce headcount or allow small finance teams to process volume that currently requires much larger teams. The financial case is elimination of transaction processing roles, not productivity improvement within those roles.
Native Integration and Security Controls
Oracle built these agents into Fusion Cloud's core transaction engine. They operate using the same security model as human transactions. When an agent processes a payment, it inherits the cost centre restrictions, approval limits, and role-based access controls that govern the user account running the agent. The agent cannot exceed budgets or thresholds that humans cannot exceed.
This is not a parallel control structure. Many ERP vendors bolt AI onto their systems and create separate approval logic for AI-assisted processes. Oracle integrated agents into the transaction core. Single source of truth. Single control framework. Agents execute within the guardrails that apply to all transactions.
AI Agent Studio: Building Custom Agents
Oracle expanded AI Agent Studio with new builder tools for customers who need agents beyond the pre-built 12. The studio includes agentic applications builder for low-code agent creation, workflow orchestration for complex multi-step processes, and ROI measurement to track cost impact. These tools allow implementation partners to customise agents for specific cost allocations, compliance rules, and integration patterns.
Custom agents will require skilled work. The low-code aspect of the builder means non-developers can perform some configuration, but most enterprise deployments will need people who understand Fusion data models, business logic, and integration patterns. Implementation partners have a significant opportunity to build and customise agents to meet specific customer requirements.
Real Scale: NHS Shared Business Services
NHS Shared Business Services is processing 355 billion pounds in annual ERP transactions on Fusion Cloud. They process 7.1 million invoices and recovered 7.4 billion pounds in automated debt collection. These are statutory audit metrics. No vendor can fabricate deployment success at this scale without auditor discovery. NHS SBS proves that Fusion agents work at scale in production.
The SaaSiQ Take
Oracle Fusion Agentic Applications represent a material shift from copilots to autonomous transaction engines. Enterprises evaluating Fusion for 2026 implementations should incorporate agent capabilities into the business case. The pre-built agents address common finance and supply chain use cases. Custom agents built through AI Agent Studio will handle variant processes. This is the direction of enterprise ERP evolution.
Mid-market finance teams should be asking Fusion partners whether agents can apply to their cost structure, business rules, and integration requirements. This is where technology investments deliver real financial impact.
Copyright SaaSiQ.ai: Intelligent Solutions for SaaS
