3 Terillium Built AI Agents will be Available in Oracle AI Agent Marketplace in Fusion 26D
Oracle is expanding its AI capabilities through the Oracle AI Agent Marketplace, where customers can access partner-built AI agents designed for specific business processes.
Terillium is bringing three Oracle AI agents to the Oracle AI Agent Marketplace with the Fusion 26D release:
- BPA Management Assistant for Procurement
- Planning Measure Expression Co-Pilot for Supply Chain Planning
- AP 3-Way Match Agent for Accounts Payable
Each agent was developed around a specific business challenge and is designed to simplify an Oracle Fusion workflow while maintaining appropriate controls and human oversight.
What Is the Oracle AI Agent Marketplace?
The Oracle AI Marketplace is a library of prebuilt AI agents, templates, and accelerators designed to help organizations adopt AI capabilities faster within Oracle applications.
Rather than building every AI capability from scratch, organizations can begin with reusable templates and AI-driven workflows that can be adapted to their business needs and ERP environments.
This approach can help organizations:
- Accelerate AI adoption
- Identify practical use cases more quickly
- Reduce development time
- Extend Oracle application functionality
- Incorporate AI into ERP implementations and optimization initiatives faster
The Oracle Marketplace is part of the broader Oracle AI Agent Studio ecosystem. Rather than requiring every organization to build an AI agent from scratch, customers can access agents developed around specific business processes and use cases. The templates in AI Agent Marketplace are validated and supported within the Fusion ecosystem.
For Oracle Fusion customers, the Oracle AI Agent Marketplace creates an opportunity to apply AI where employees spend significant time:
- Searching for information
- Navigating through applications
- Validating data
- Performing repetitive tasks
- Managing exceptions
Terillium’s three AI agents in Marketplace demonstrate this approach across procurement, supply chain planning, and accounts payable.
Terillium AI Agents Coming to Oracle Fusion 26D
The Oracle AI Agent Marketplace gives Fusion customers access to partner-built agents for specialized business processes. Terillium’s initial Marketplace offerings focus on three areas:
These agents have been validated and approved by Oracle for the Oracle AI Agent Marketplace for the Fusion 26D release.
Terillium’s BPA Management Assistant
Procurement teams often need to search, review, and update Blanket Purchase Agreements (BPAs). Finding the right agreement and making changes can require navigating through multiple areas of an ERP system.
The Terillium BPA Management Assistant provides a conversational way for buyers to search and update BPAs through Oracle AI Chat without navigating through the Fusion user interface.
What Can the BPA Management Assistant Do?
The agent allows buyers to:
- Search BPAs by agreement number
- Search by supplier
- Search by item or item description
- Update item prices
- Add new item lines
- Add quantities
- Maintain conversational context from one request to the next
For example, a buyer could ask the agent: “Look up BPA 460013.” Or “Show me all BPAs for supplier X”. The agent can also handle requests such as updating an item’s price or adding a new item to an existing BPA.
The agent originated from a real business use case involving a manufacturing and distribution client, where the Terillium team was working with multiple inventory organizations and encountered challenges around managing pricing through BPAs and maintaining an efficient purchase order process.
A primary goal of the agent was to reduce the number of clicks required to complete BPA-related tasks. As a partner-built solution, the BPA Management Assistant will give Oracle Fusion procurement teams access to this type of specialized capability through the Oracle AI Agent Marketplace.
How the Procurement AI Agent Works
The user submits a request in natural language. The agent classifies the user’s intent and automatically calls the appropriate Fusion REST API to complete the request. It uses Oracle Fusion REST APIs and business objects to work with the underlying BPA data.
Human approval is required for write operations, allowing users to review changes before they are made.
Key benefits:
- Less manual navigation
- Faster BPA searches
- Easier procurement updates
- Conversational access to BPA information
- Human oversight for changes
Terillium’s Planning Measure Expression Co-Pilot
Creating and troubleshooting planning measure expressions can be challenging for Oracle Supply Chain Planning users. Expression syntax is unforgiving, and incorrect measure names or syntax can cause an expression to fail or produce incorrect results.
The Terillium Planning Measure Expression Co-Pilot is designed to help users build, understand, troubleshoot, and validate planning expressions within Oracle Fusion.
What Can the Supply Chain Planning AI Agent Do?
The agent classifies each request into one of six intents and routes it to the appropriate workflow. Users can use the agent to:
- Build new measure expressions from natural-language descriptions
- Explain existing measures and expressions
- Troubleshoot expressions that fail validation
- Check whether a measure exists in the system
- Answer planning-related questions
- Fix expressions and provide output ready to use
The agent validates every measure name against the live Planning Measures REST API, helping ground its responses in the customer’s actual Fusion environment. This makes the AI Agent particularly useful for planning teams that need to create custom measures without relying on IT for every expression-building task.
AI-Powered Oracle Supply Chain Planning Workflows
Rather than simply asking an AI model to generate an expression, Terillium designed the agent around validation and deterministic workflows.
Key design elements include:
- 24-node Workflow with deterministic intent routing
- Validate-then-build approach to help prevent fabricated measure names
- Structured prechecks for troubleshooting
- Per-node audit traceability
- Approximately 78% lower token consumption on the explain-existing path compared with v1
- Built natively on Oracle AI Agent Studio
- Zero data egress
- No third-party LLM provider
- No additional licensing
The agent was developed in response to limitations identified with the seeded Planning Measure Expression Assistant and was designed to provide a more structured workflow for planning users.
As an agent being brought to the Oracle AI Agent Marketplace, the AI agent provides Fusion Supply Chain Planning users with access to a purpose-built workflow rather than a generic AI assistant.
Terillium’s AP 3-Way Match Agent
Accounts Payable teams can spend significant time comparing supplier invoices with purchase orders and receipts to determine whether an invoice can be processed.
The Terillium AP 3-Way Match Agent automates this reconciliation process within Oracle Financials Cloud Payables.
How Does the AP 3-Way Match Agent Work?
The traditional three-way match compares:
Supplier invoice → Purchase order → Receipt
The Terillium agent evaluates the invoice line by line, checking information such as:
- Quantity
- Price
- Amount
- Currency
Invoices can enter the agent through an invoice number, line information, or a PDF attachment. The agent retrieves the relevant PO and receipt information from Fusion and applies configurable tolerances to determine the appropriate result.
Each line is evaluated as a PASS, WARN, or FAIL, allowing the AI agent to distinguish between invoices that can move forward and those that require additional attention
This brings a specialized AP automation use case to the Oracle AI Agent Marketplace, giving Fusion customers another way to streamline repetitive financial processes.
How the AP 3-Way Match Agent Handles Invoice Exceptions
The workflow includes five possible outcomes:
- MATCHED
- PARTIAL_MATCH
- MISMATCH
- NO_PO_FOUND
- NO_RECEIPT_FOUND
Each path has its own audit trace, notification template, and return workflow, so exceptions aren’t simply dropped from the process.
The AI Agent can then generate a line-level report and direct users back into Fusion to determine how an exception should be handled.
This approach can be useful for:
- High-volume AP teams where clean invoices are still manually reviewed
- Shared service centers that need consistent, auditable matching decisions
- Finance organizations with strict controls around AP automation
- Organizations dealing with invoices where line-level discrepancies may not be apparent from the total
Human Approval in the AP 3-Way Match Process
The AP 3-Way Match Agent is designed to automate the repetitive reconciliation work without removing people from important financial decisions.
The agent operates with read-only access to Oracle Fusion, while final invoice posting and payment release remain human actions within the standard Fusion process.
This allows AP teams to spend less time manually comparing information and more time addressing exceptions and higher-value work.
Potential Business Impact of AP Automation
Terillium’s model estimates that for a customer processing 10,000 invoices per month, automating the matching step could potentially reclaim approximately 1,200 AP hours per month. At a fully loaded AP rate of $30 per hour, that represents approximately $432,000 in modeled annual labor savings.
This is an illustrative model, not a guaranteed savings figure. Actual results will vary based on invoice volume, existing processes, exception rates, labor costs, and implementation.
How Terillium is Applying AI to Oracle Fusion
The development of these agents for Oracle AI Agent Marketplace represents one way Terillium is helping organizations move from conversations about AI to practical AI applications within Oracle Fusion.
Terillium is also working directly with customers to build custom AI agents tailored to their unique business processes, operational requirements, and Oracle applications. This combination of reusable marketplace capabilities and customer-specific innovation helps organizations accelerate AI adoption while addressing their specific business needs.
Terillium’s Oracle AI Strategy Workshop
Knowing where to apply AI is just as important as knowing what the technology can do. Terillium’s Oracle AI Strategy Workshop helps organizations take a structured approach to identifying and prioritizing AI opportunities.
During the workshop, Terillium works with leadership teams to:
- Identify operational pain points where AI could create meaningful improvements
- Evaluate potential AI initiatives based on ROI, operational impact, and implementation complexity
- Consider data, governance, and security requirements
- Align AI opportunities with broader business goals
- Develop a phased roadmap for moving from exploration to implementation
Organizations leave the workshop with a clearer understanding of where AI can drive measurable value, a high-level view of potential ROI, and a roadmap for moving forward.
Whether supporting a new ERP implementation or helping existing Fusion managed services customers evaluate emerging Oracle Fusion AI capabilities, Terillium combines deep ERP expertise with practical AI strategies to help organizations achieve measurable business outcomes.se with practical AI strategies to help organizations achieve measurable business outcomes.
