Oracle EPM Agentic AI Strategy and Roadmap: Finance Moves from Assisted Work to Autonomous Operations

Finance teams are entering a new operating model. The next phase of enterprise performance management is not only about faster reporting, better dashboards or automated calculations. It is about AI agents that can interact with finance data, explain variances, guide workflows, support reconciliations, generate narratives and eventually move finance teams toward more autonomous operations. 

Oracle’s AI Agentic Strategy & Roadmap for EPM presents a clear direction: bring intelligent agents directly into EPM, expose governed finance actions through Agentic Assistants and allow interoperability with broader AI platforms such as ChatGPT, Claude and OCI AI Agent Platform through REST APIs. The session was presented by Oracle EPM Product Management on May 28, 2026.  

Oracle includes a safe harbor statement that roadmap items are for information purposes only and that development, release, timing and pricing may change at Oracle’s discretion.  

The Three Pillars of Oracle EPM AI Agents 

Oracle’s EPM AI strategy is organized around three pillars: 

• Agentic Experiences  
• Agentic Assistants  
• Interoperability with AI Platforms  

This structure shows that Oracle is not treating AI as a single chatbot layer. It is building a broader EPM AI architecture that supports embedded experiences, governed assistant capabilities and external AI platform interoperability.  

Agentic Experiences: Intelligent Agents Directly Inside EPM

Agentic Experiences bring native AI agents directly into the EPM application experience. Oracle positions these as intelligent agents that allow users to interact with planning, financial close and reporting through natural conversation. 

The roadmap includes examples such as: 

• Planning Agent  
• Reconciliation Agent  
• Reporting Agent  

These agents are designed to support conversational assisted actions, guided finance workflows and, over time, autonomous finance operations.  

Oracle describes the progression as increasing autonomy: 

Conversational → Guided Workflows → Autonomous 

This progression is governed by an enterprise governance layer that includes policy documents, financial governance rules, approval workflows, transparency and auditability.  

Why Agentic Experiences Matter 

Traditional EPM users often move between forms, dashboards, reports, grids, journals, reconciliations and spreadsheets. They know what question they want answered, but they must manually navigate the system to find the right form, report or data view. Agentic Experiences change that interaction model. 

Instead of navigating through multiple screens, users can ask natural-language questions, retrieve finance data, visualize trends, explain variances and move closer to action within the same workflow. For finance teams, this can reduce time spent searching, clicking and interpreting static outputs. It can also help shift EPM from a reporting system into an interactive decision layer. 

Conversational AI in EPM 

Oracle’s roadmap shows conversational AI in both browser and Smart View experiences. The visual on page 6 shows natural-language interaction inside the EPM browser interface and within Oracle Smart View, allowing users to work with finance data through conversation rather than only traditional forms or reports.  

Oracle also shows hybrid and full-screen interactions with tables and visualizations, where users can interact with dashboards, reports and charts through a conversational side panel or a fuller analysis view. This is important because finance users do not only need answers in text. They need tables, charts, comparisons, explanations and data-backed insights. 

Planning Agent: From Static Planning to Interactive Financial Reasoning 

The Planning Agent is one of the major Agentic Experience examples in Oracle’s roadmap. Oracle describes the Planning Agent as providing conversational access to plans, actuals and insights. Users can drill into data, explain variances and surface drivers without navigating forms or reports. It is also designed to accelerate insight-to-action across the FP&A workflow by moving analysts from static report consumption to interactive financial reasoning.  

The Planning Agent includes capabilities such as: 

• Conversational access to planning data  
• Text, chart and table responses  
• Data Exploration Assistant  
• What-if Prediction Assistant  
• Insight Exploration Assistant  
• What-if analysis against predictive forecasts  

This helps FP&A teams ask business questions directly and receive data-backed explanations, rather than manually building every analysis from scratch. 

Planning Agent Key Capabilities 

Oracle highlights two major capability areas for the Planning Agent:

Daily P&L Forecasting

Daily P&L Forecasting includes: 

• Chart of account alignment with ERP  
• Application launch from ERP  
• Integration with analytical views  
• Financial Planning Application 2.0  
• Financial and operational drivers from Fusion  
• Drill-through from EPM to Fusion sources 

Conversational and Agentic AI

The Planning Agent also supports: 

• Conversational AI  
• Text, charts and insights  
• Hybrid and full-screen interaction  
• Intent classification  
• Semantic search  
• Built-in context through EPM Genie  
• Root cause analysis with agent-to-agent interactions  
• Extensibility through AI Agent Studio  

This shows Oracle’s direction toward planning experiences that are deeply connected to Fusion data and process, not isolated forecasting screens.  

Financial Planning 2.0: Redesigned for Agentic Experiences 

Oracle’s roadmap introduces Financial Planning 2.0, described as redesigned for agentic experiences. 

Key capabilities include: 

• Dimensions based on chart of account segments  
• Seamless integration and drill-through to Fusion Analytical Views  
• Multi-cube design  
• Revenue, Expense and Financial Statement cubes  
• Extended dimensionality  
• Blended planning and forecasting methods  
• Trend, driver, prediction, smart drivers and direct input  
• Seeded application content  
• Driver-based models for revenue and expense  
• Forms, dashboards and reports  
• Insights for automated analysis  
• Auto Predict and Advanced Prediction models  
• IFRS 18 financial statements for Management Performance Measures  
• EBITDA financial statement  

This is a significant direction for finance transformation because planning becomes more connected to ERP structure, operational drivers and AI-assisted forecasting.  

Narrative Reporting Agent: Making Reports Interactive 

Oracle also highlights the Narrative Reporting Agent, which is designed to provide conversational access to reports. 

The roadmap positions this agent as allowing users to: 

• Perform ad hoc analysis on reports  
• Add calculated columns  
• Investigate conditional exceptions  
• Turn reports from static artifacts into interactive tools  
• Generate narratives and insert them as notes  
• Modify generated notes  

Oracle presents a phased release target: 

• 26.07: Conversational access to reports for static analysis  
• 26.09: Ad hoc querying, causality investigation and generative narrative as notes  
• 1H CY27: Report authoring  

These items are roadmap targets and should be treated under Oracle’s safe harbor guidance.  

Reports Gen AI Agent: Three-Phase Progression 

Oracle’s report agent roadmap is structured into three phases: 

Phase 1: Understand 

Static analysis and report discovery. Users can ask direct questions on a selected report grid, launch the right report from Ask Oracle in the library and perform focused Q&A on existing report content. 

Phase 2: Investigate 

Dynamic investigation and causal analysis. Users can investigate drivers behind report values and variances, generate focused analysis views with ranking and filtering, and create reusable grid outputs from the analysis flow. 

Phase 3: Edit and Author 

Natural-language editing and guided authoring. Users can edit common grid structures through natural language and prepare for broader creation of report objects and insights. Oracle describes the outcome as a single GenAI experience that moves from answering questions to investigating issues and then helping users edit and capture report insights.  

Reconciliation Agent: Toward Continuous Reconciliation 

The Reconciliation Agent is one of the most important developments for financial close teams. Oracle describes the Reconciliation Agent as using AI-powered automation to provide continuous reconciliation processing throughout the period, helping remove the end-of-period fire drill. The agentic flow moves from helping users gather information to support a balance, to automatically gathering and closing autonomously over time. Oracle notes that users may gradually shift from doing the activity to monitoring the activity.  

For finance leaders, this is a major shift. Instead of reconciliation being concentrated around period-end pressure, AI-assisted reconciliation can support a more continuous close model. It can help teams prepare reconciliations earlier, surface issues sooner and reduce manual effort during the close window.

Agentic Assistants: Governed Finance Actions

The second pillar is Agentic Assistants. Oracle describes Agentic Assistants as composable, governed finance capabilities that Fusion AI agents can assemble into customer-specific workflows. These assistants automate finance work without requiring users to start from a blank prompt or rebuild business logic from scratch.  

Examples shown include: 

• EPM Assistant  
• Consolidation Journals Assistant  
• Account Reconciliation Assistant  
• Narrative Reporting Assistant  

Oracle’s sample workflow shows a user exploring a balance sheet variance, reviewing journal status, checking reconciliation status and reviewing a report using different assistants across the EPM process.  

Agentic Assistants Across EPM Business Processes 

Oracle’s roadmap lists 33 Agentic Assistants across EPM business processes and release waves. 

These include assistants for: 

• Platform  
• Narrative Reporting  
• EPCM  
• Tax  
• FCC  
• Account Reconciliation  
• Task Manager  
• Supplemental Data  
• Enterprise Journals  
• Planning  
• EDM  

Examples include EPM Assistant for users and administrators, Narrative Reporting Assistant, Account Reconciliation Assistant, Transaction Matching Assistant, Task Manager Assistant, Enterprise Journals Assistant, Long Range Planning Assistant and EDM Request Authoring Assistant. Oracle positions these assistants as serving users, administrators and implementers, helping automate a broad range of tasks across Cloud EPM and simplify operations.  

How Agentic Assistants Are Used 

Oracle shows a three-step assistant process: 

• Download the AI Studio import from the EPM environment.  
• Import the Assistant .json file into AI Studio.  
• Use the assistant.  

This means Agentic Assistants are not only conceptual roadmap items. They are designed to be operationalized through AI Studio workflows where EPM capabilities can be exposed and assembled.  

Interoperability with AI Platforms

The third pillar is Interoperability with AI Platforms. Oracle shows that EPM can use existing REST APIs in the context of AI interoperability and can be deployed with platforms such as Claude, ChatGPT and OCI AI Agent Platform. Oracle notes that this approach is flexible but requires custom development, with REST endpoint improvements on the roadmap.  

This is important for organizations building broader AI strategies beyond a single enterprise application. Finance AI may need to interact with ERP, EPM, data warehouses, collaboration tools, external LLMs and enterprise AI platforms. Oracle’s interoperability model supports this direction while still recognizing that custom development effort is required. 

Smart View CoPilot Agent 

Oracle also highlights the Smart View CoPilot Agent, described as a Dynamic Visualization Assistant. 

Capabilities include: 

• Auto-generating appropriate charts and visuals during data exploration  
• Toggling between visualization and charting formats  
• Expected availability in Web and Oracle Smart View interfaces  
• Embedded EPM Chat or use as a CoPilot Agent  

For finance teams that live in Excel and Smart View, this is especially relevant. It brings conversational visualization and analysis closer to the spreadsheet experience where many planning and reporting users already work.  

Oracle EPM Agentic AI Release Strategy 

Oracle’s roadmap slide shows release strategy across 2026 and 2027, covering interoperability, Agentic Assistants and Agentic Experiences. 

The roadmap includes: 

• Smart View CoPilot Agent for conversational chart building and automated visuals  
• Agent-ready business actions and Modeling Agent  
• Assistants across Platform, Narrative Reporting, EPCM, Tax, FCC, AR, Planning and EDM  
• Tax RTP Adjustment  
• EDM Conversational Assistant  
• FCC capabilities such as rules creation, close processing and ownership management  
• Planning Agent with data exploration, insight exploration and predictive modeling  
• Reporting Agent with conversational access, ad hoc querying and causality investigation  
• Planning Agent for ERP with conversational drill and root cause analysis  
• Reconciliation Agent with preparation assistant and continuous reconciliations  
• Fusion AI Agent Studio capabilities including agent-to-agent, third-party data and Slack/Teams integration  

These are roadmap items and should be considered under Oracle’s safe harbor statement.  

Additional AI Enhancements Coming to EPM 

Oracle also lists several additional AI enhancements coming across predictions, insights, Financial Consolidation & Close, Account Reconciliations and Predictive Cash Forecasting. 

Predictions 

Planned enhancements include new regression models in Advanced Predictions, driver support at different POV intersections, improved Auto Predict accuracy logic, non-negative prediction options, allocation rule/template support for dynamic parent predictions, real-time what-if predictions for Advanced Predictions and multi-member target selection.  

Insights 

Planned enhancements include insight summarization for account members, in-context insights, periodic movement variance insight type, root cause analysis for EPM cubes, enhanced insights in Smart View, GenAI support for non-English languages, insights integration to Narrative Reporting, attributes in insights and expanded history in insight graphs.  

Financial Consolidation & Close 

The roadmap includes Validations Hub and Script Narration. 

Account Reconciliations 

The roadmap includes many-to-one and one-to-many match predictions, pre-submission analysis for potential rejection and NLP filtering for reconciliations. 

Predictive Cash Forecasting 

The roadmap includes drill to Fusion Analytical Views in ERP, PCF for Fusion ERP to be generally available in 26B and bank statements for cash positioning.  

Validations Hub: AI-Supported Close Control 

In the appendix, Oracle highlights Validations Hub as supporting out-of-the-box and configurable validations and financial statement analysis. The slide describes Validations Hub as automating a key control in consolidation and close activities. Use cases can range from common balancing items to disclosures and management reporting detail.  

This is relevant because AI-driven finance still needs strong control frameworks. Validations, auditability and exception management remain central to responsible EPM transformation. 

AI Agent Options for Planning 

Oracle compares three AI agent options using Planning as an example: 

Agentic Experiences 

These include the built-in Planning Agent with Data Exploration Assistant, Insight Exploration Assistant and Prediction Assistant. They are embedded as conversational AI and can return text, charts, insights and predictions. Oracle describes this as easier to turn on and use with minimal consulting effort.  

Agentic Assistants 

These include the EPM Assistant, partially built in, with templates and business objects. They are deployed through AI Agent Studio and may require AI Agent Studio expertise and prompt design.  

Interoperability with AI Platforms 

This uses REST APIs and can work with Smart View CoPilot or external AI platforms. It is flexible but requires higher consulting effort and agentic expertise with any LLM. This comparison helps organizations decide where to start based on maturity, use case complexity and available expertise. 

Why This Matters for CFOs and Finance Transformation Leaders 

Oracle’s EPM Agentic AI roadmap shows a clear movement from manual finance work toward AI-assisted and increasingly autonomous finance operations. For CFOs, this is not only a technology update. It affects how finance work is designed. 

The opportunity is to: 

• Reduce time spent navigating reports and forms  
• Improve speed of variance analysis  
• Support predictive planning and scenario simulation  
• Strengthen narrative reporting and management commentary  
• Move reconciliation closer to continuous processing  
• Create governed finance workflows using AI assistants  
• Extend EPM intelligence into external AI platforms  
• Improve close, planning and reporting productivity  

But success will depend on readiness. Data quality, process governance, security, approval workflows, metadata design, reporting structures and user adoption will matter as much as the AI tools themselves. 

How NexInfo Can Help 

NexInfo helps organizations prepare for Oracle EPM Agentic AI by aligning technology adoption with finance process transformation. 

NexInfo can support: 

• Oracle EPM AI readiness assessment  
• Planning Agent readiness and use case design  
• Reconciliation Agent adoption roadmap  
• Narrative Reporting Agent readiness  
• Agentic Assistant planning and AI Studio enablement  
• EPM process and data readiness review  
• Financial planning modernization  
• Continuous reconciliation roadmap  
• Reporting and narrative automation strategy  
• Smart View CoPilot readiness  
• REST API and AI interoperability planning  
• EPM governance and control framework design  
• User training and adoption support  
• Release roadmap assessment and implementation planning  

NexInfo helps finance teams move from interest in AI to practical, governed and measurable EPM adoption. 

Conclusion 

Oracle’s EPM Agentic AI strategy signals a major shift in enterprise performance management. Planning, reporting and reconciliation are moving from static workflows toward conversational, guided and eventually autonomous finance operations. 

The roadmap includes Planning Agent, Reconciliation Agent, Narrative Reporting Agent, Agentic Assistants, Smart View CoPilot, REST API interoperability, Financial Planning 2.0 and additional AI enhancements across predictions, insights, consolidation, reconciliation and cash forecasting. 

For finance leaders, the message is clear: AI in EPM is no longer limited to forecasting or anomaly detection. It is moving toward governed finance agents that can assist, explain, recommend and act within enterprise controls. 

NexInfo helps organizations prepare for this future with Oracle EPM expertise, AI readiness planning, process governance, integration capability and implementation support. 

FAQ

What is Oracle EPM Agentic AI?

Oracle EPM Agentic AI refers to Oracle’s roadmap for intelligent AI agents, assistants and interoperability capabilities across Enterprise Performance Management. It includes agentic experiences inside EPM, governed finance assistants and the ability to connect EPM with external AI platforms through REST APIs.

What are the three pillars of Oracle EPM AI Agents?

Oracle’s strategy includes three pillars: Agentic Experiences, Agentic Assistants and Interoperability with AI Platforms. These pillars support native EPM agents, governed finance capabilities and external AI platform connectivity. 

What is the Oracle Planning Agent?

The Oracle Planning Agent is an EPM AI agent that provides conversational access to plans, actuals and insights. It supports data exploration, what-if prediction, insight exploration, variance explanation and interactive financial reasoning. 

What is the Oracle Reconciliation Agent?

The Oracle Reconciliation Agent is designed to support AI-powered continuous reconciliation processing throughout the period. Oracle describes the agentic flow as moving from helping users gather information to supporting automated gathering and autonomous closing over time. 

What is the Oracle Narrative Reporting Agent?

The Oracle Narrative Reporting Agent provides conversational access to reports, ad hoc analysis, conditional exception investigation, generative narrative creation and report authoring capabilities over phased roadmap releases. 

What are Agentic Assistants in Oracle EPM?

Agentic Assistants are composable, governed finance capabilities that Fusion AI agents can assemble into customer-specific workflows. Examples include EPM Assistant, Account Reconciliation Assistant, Consolidation Journals Assistant and Narrative Reporting Assistant. 

Does Oracle EPM work with ChatGPT or Claude?

Oracle’s roadmap includes interoperability with AI platforms using REST APIs. The deck mentions deployment with Claude, ChatGPT and OCI AI Agent Platform, while noting that this is flexible but requires custom development. 

What is Smart View CoPilot Agent?

Smart View CoPilot Agent is a Dynamic Visualization Assistant that can auto-generate charts and visuals during data exploration and allow users to toggle between visualization and charting formats in Web and Oracle Smart View interfaces. 

What is Financial Planning 2.0 in Oracle EPM?

Financial Planning 2.0 is a redesigned planning model for agentic experiences. It includes chart-of-account based dimensions, Fusion Analytical Views integration, multi-cube design, blended forecasting methods, seeded content, insights, Auto Predict and Advanced Prediction models. 

How can NexInfo help with Oracle EPM Agentic AI?

NexInfo can help organizations assess AI readiness, identify EPM agent use cases, prepare planning and reconciliation workflows, design governance models, support AI Agent Studio adoption, plan integrations and guide finance teams through Oracle EPM AI transformation.