AI adoption in finance is moving into a more practical phase. CFOs, FP&A leaders and EPM administrators are no longer asking whether AI can support planning, forecasting and analysis. The bigger question is how finance teams can use AI securely, with the right access controls, explainability, governance and operational ownership.

Oracle Cloud EPM Intelligent Performance Management is evolving in that direction. Oracle’s October 2025 Cloud Customer Connect material focuses on two important areas: the new IPM Manage granular role and the forward roadmap for Insights, Advanced Predictions, GenAI and Planning Agent capabilities.

The direction is clear: Oracle is making AI in EPM more accessible to finance users while preserving security, control and administrative oversight.

Why IPM Matters in Oracle Cloud EPM

Intelligent Performance Management, or IPM, is Oracle’s AI and machine learning layer for EPM. It helps finance teams monitor business performance, predict outcomes, automate analysis and engage with planning data in more intelligent ways.

Oracle positions AI in EPM across four business questions:

Automate: What processes can be fully automated?
Predict: How can finance better anticipate the future?
Monitor: What are the most material exceptions to be aware of?
Engage: How can GenAI empower finance objectives?

This framing is useful because it places AI inside real finance work. IPM is not only about producing a prediction. It is about helping finance teams identify exceptions, explain drivers, improve forecast confidence and accelerate planning decisions.

Oracle Cloud EPM AI Maturity Is Increasing

Oracle’s AI innovation path in Cloud EPM has progressed from predictive AI in 2016 to AutoPredict, Insights, Predictive Cash Forecasting, Narrative Generation, Modeling Agent, Narrative Summarization and skills-based AI agents and agent teams.

The material also notes that more than 1,600 Oracle Cloud EPM customers, approximately 20%, had at least one AI feature enabled in production. This shows that EPM AI adoption is no longer limited to early experimentation. More customers are beginning to operationalize AI features inside production finance environments.

The New IPM Manage Role: Secure Self-Service for Finance Teams

One of the most important updates is the IPM Manage granular role, released in 25.10 and available for Enterprise customers. Oracle describes the purpose clearly: empower finance teams to configure and run IPM securely without IT bottlenecks. This is a significant governance improvement.

Historically, finance users may have depended on Service Administrators to configure certain AI and IPM-related jobs. With IPM Manage, selected business users can create and manage jobs independently while still respecting existing EPM security.

The role supports configuration and management for:

Insights
Auto Predict
Advanced Predictions

For finance teams, this can reduce dependency on IT and allow analysts, planners and FP&A users to work more directly with AI capabilities.

Why Granular AI Roles Matter

AI governance becomes harder when access is too broad or too centralized. If only administrators can configure AI jobs, finance teams move slowly. If too many users receive broad administrative rights, the control model becomes risky. The IPM Manage role creates a middle path. It gives finance users controlled self-service while allowing administrators to retain oversight.

Oracle highlights five key benefits:

Granular role control for IPM job configuration
Assignment to individual users or user groups
Member-level security honored
Jobs accessible only to the creator, with administrators retaining full visibility
Finance teams can manage IPM configurations without IT dependency
This is the right direction for enterprise AI adoption: democratized use, but governed execution.

Individual User and Group Assignment Use Cases

Oracle identifies two practical use cases for the IPM Manage role. The first is an individual user scenario. A business user with IPM Manage can create and manage jobs independently, reducing reliance on Service Administrators for setup. This helps planners and analysts configure Insights, Auto Predict and Advanced Prediction jobs directly.

The second is group assignment. The role can be assigned in bulk to an FP&A or functional group, allowing every member of that group to create and manage their own jobs. This simplifies administration while enabling teams to operate more efficiently. For larger finance teams, group assignment can be especially useful because AI job ownership can align with functional responsibilities.

Security Assurance Remains Central

Oracle emphasizes that all activities remain governed by existing EPM security. Users can only access and run jobs for dimension members they are authorized to access. This is critical. AI should not bypass existing application security. If a finance user is restricted from certain entities, accounts, business units or dimensions, AI job access must honor those same restrictions.

For organizations with sensitive planning, compensation, workforce, profitability or financial data, this security model helps reduce risk while expanding AI usage.

IPM Roadmap: A Broader Finance AI Agenda

Oracle’s IPM roadmap focuses on three major areas:

Insights
Predictive Planning and Advanced Predictions
GenAI and Planning Agent

This roadmap shows that Oracle is investing in both analytical depth and user experience. The future of IPM is not just better algorithms. It is better explanations, better workflow integration, more context-aware insights and AI agents that help users make faster decisions.

Advanced Predictions: From Forecast Output to Explainable Forecasting

Forecasting is only useful when finance users trust the prediction. A number without explanation can create more questions than answers. Oracle’s roadmap for Advanced Predictions includes feature engineering, feature importance, feature selection, what-if predictions, multi-member target selection, drivers at different POV intersections, allocation rules and templates for dynamic parent predictions, improved Auto Predict accuracy logic, non-negative predictions and Smart View parity for Interactive Predictive Planning on forms.

The direction is clear: predictions are becoming more explainable, configurable and aligned with finance decision-making.

Feature Importance: Building Trust in Machine Learning Predictions

Oracle highlights explainable predictions through Feature Importance. Feature Importance helps users understand which business drivers impact forecasts the most and compare the relative contribution of each driver. It also supports fitted-line visualization of historical data and model estimates.

This is valuable because finance users need to challenge assumptions, validate drivers and understand why a model produced a forecast.

Feature Importance can help teams:

Identify key business levers
Understand forecast drivers
Compare driver contribution
Improve forecast transparency
Support auditability
Build confidence in ML predictions
Enable better finance review conversations

This moves AI forecasting from “black box output” to explainable planning intelligence.

Intelligent Feature Engineering

Feature engineering is the process of preparing data for machine learning by transforming existing features or creating new ones to improve model performance. Oracle highlights support for time-based features, lag effects, significant lag using autocorrelation and aggregate transformations such as rolling mean and rolling median. This is highly relevant for FP&A.

Business drivers often do not affect results immediately. Marketing spend may influence revenue after a delay. Hiring may affect productivity over several periods. Pricing changes may influence volume with a lag. By supporting lag effects and rolling averages, Advanced Predictions can better represent the real-world timing of business impact.

Intelligent Feature Selection

Oracle also highlights Feature Selection, which chooses the most relevant variables from a dataset while considering how they interact with the target variable. The goal is to determine whether a feature adds value and automatically add or remove it from the prediction model.

Too many low-value variables can weaken predictions. Feature Selection helps reduce noise, avoid overfitting, improve processing time and make models easier to interpret.

Oracle identifies algorithms such as:

Random Forest Regressor
Correlation coefficients
Lasso Regression
Elastic Net
Ridge Regression
Auto Mode

For finance users, this means the model can focus on meaningful drivers instead of overwhelming the forecast with every available variable.

Intelligent Data Selection

Oracle’s short-term roadmap includes Intelligent Data Selection. This capability uses only series with valid actuals in the driver slice for prediction. Series with missing actuals are automatically ignored, helping prevent job failures. Oracle gives an example where future series exist for a discount percentage driver, but because actual data is missing, those series are excluded from prediction.

This is a practical improvement for real-world planning environments. Data is rarely perfect. Intelligent Data Selection can help forecasting jobs become more resilient by avoiding avoidable failures caused by incomplete driver data.

Dynamic Calc in Advanced Predictions

Oracle also identifies dynamic calculation support in Advanced Predictions as part of the short-term roadmap. Dynamic calc predictions were previously available for Predictive Planning and are now being extended to Advanced Predictions. This allows business users and FP&A teams to make predictions at parent-level or dynamic calc members without depending on granular leaf-level data. This is important when leaf-level data is sparse, inconsistent or too detailed for meaningful prediction.

For example, users may want to predict at a total travel and entertainment level rather than at every detailed subcategory. Oracle notes use cases where leaf member data may not be suitable for prediction or users may not want insights on very detailed accounts. This creates more flexibility for configuring AI around actual business planning needs.

Insights Roadmap: Making Exceptions More Contextual

Oracle’s Insights roadmap includes several enhancements, such as periodic movement variance insight type, in-context insights, root cause analysis, integration with Narrative Reporting, enhanced Insights in Smart View, attributes in Insights, exclude options in slice definitions and two years of history in insight graphs.

This matters because insights should appear where finance users work. They should not remain hidden in a separate analytical layer that only power users access.

In-Context Insights

Oracle describes in-context insights as a way to make insights available to a wider user base by surfacing them on forms and dashboards. The goal is to accelerate AI adoption and make the Insights engine more intelligent by adapting calculations based on custom settings for data intersections. This is a strong usability improvement.

Instead of asking users to leave their planning form or dashboard to search for insights, the system can bring relevant insights directly into the workflow. This can increase adoption because insights become part of normal planning review.

Periodic Movement Variance Insight

The Periodic Movement Variance Insight is designed to monitor a specific Account member, such as a measure, KPI or metric of interest. It triggers an insight when the change in the recent period is materially significant compared to a benchmark based on the previous period.

Oracle notes that this capability will be relevant across Planning, Strategic Workforce Planning, Financial Close, Tax, Profitability and Cost, and other areas. For finance teams, this is valuable because many important business issues show up as unusual period movement. Detecting those movements automatically can help teams focus attention where it matters most.

Root Cause Analysis

Oracle’s roadmap includes Root Cause Analysis for detected insights such as anomalies and forecast variances. It is designed to generate a structured breakdown of contributors, potential correlations and causal factors.

The roadmap shows a staged direction:

Driver and contributor analysis at cube level
Causal and correlation analysis at cube level
Causal and correlation analysis using Fusion transactions
Causal and correlation analysis using operational models
Causal and correlation analysis using third-party sources through AI Agent Studio

This is one of the most important roadmap themes because finance teams do not only need to know that an anomaly happened. They need to know why it happened. Root Cause Analysis can help FP&A teams move from exception detection to decision support.

GenAI Roadmap: Better Summaries and Planning Interaction

Oracle’s GenAI roadmap includes insight summarization for Account members, GenAI support for non-English languages and the Planning Agent. A specific enhancement allows multi-insight summaries to support the Account dimension. Previously, users could select multiple insights across dimensions excluding Account in a parent-child context and receive a consolidated summary. Now, users can select insights across multiple account members in a parent-child context to generate a trend summary.

This is valuable because finance teams often analyze performance by account hierarchy. Being able to summarize insights across account members can improve operating expense review, revenue analysis, margin commentary and management reporting.

Planning Agent: AI Embedded into FP&A Workflows

Oracle defines the Planning Agent as an AI-powered agent embedded into EPM workflows to accelerate planning, analysis and decision-making for FP&A users. Its core purpose is to move beyond static reports and manual analysis and deliver proactive, on-demand insights for finance users. The reason for this direction is clear. Enterprises need faster decision cycles, better scenario resilience and automated insight generation in a dynamic business environment.

For FP&A teams, the Planning Agent represents the next stage of planning experience. Instead of manually reviewing forms, dashboards and reports, finance users can receive guided assistance, analysis and decision support inside the planning workflow.

Business Impact for CFOs and FP&A Leaders

The IPM roadmap has several practical implications for finance leadership.

AI Moves Closer to Business Users

The IPM Manage role allows finance teams to configure AI jobs without relying entirely on administrators.

Forecasts Become More Explainable

Feature Importance, Feature Engineering and Feature Selection improve confidence in machine learning predictions.

Planning Becomes More Resilient

Intelligent Data Selection helps reduce failed prediction jobs caused by incomplete driver data.

AI Becomes More Contextual

In-context insights bring intelligence into forms and dashboards where users already work.

Root Cause Analysis Strengthens Decision Support

Finance teams can move beyond detecting anomalies toward understanding contributors and possible causal factors.

Planning Agent Changes the FP&A Experience

AI agents can help finance teams move from static analysis to proactive, on-demand planning support.

How NexInfo Can Help

NexInfo helps organizations adopt Oracle Cloud EPM IPM capabilities with the right strategy, governance and implementation roadmap.

NexInfo’s enterprise delivery model is supported by ISO 9001 Quality Management and ISO 27001 Information Security certifications, helping organizations strengthen delivery discipline, process quality and information security governance. NexInfo has also received the AI-Enabled Workforce Excellence Award at the 1st Annual Long Beach Business AI Summit, reinforcing its focus on practical AI adoption across workforce enablement, enterprise systems and operational transformation.

NexInfo can support organizations with:

Oracle Cloud EPM IPM readiness assessment
IPM Manage role planning and governance
Insights configuration and adoption
AutoPredict enablement
Advanced Predictions design
Feature driver and model readiness review
Forecast explainability planning
Root Cause Analysis readiness
In-context insight adoption strategy
Planning Agent readiness roadmap
EPM security and role review
User training and change management
Managed support for continuous optimization

NexInfo helps finance teams move from AI interest to controlled, secure and measurable AI adoption in Oracle Cloud EPM.

Conclusion

Oracle Cloud EPM IPM is becoming more practical, more governed and more embedded into finance workflows. The IPM Manage role gives finance teams secure self-service control over AI job management. Advanced Predictions are becoming more explainable through feature importance, feature engineering and feature selection. Insights are becoming more contextual through forms, dashboards, periodic movement variance and root cause analysis. GenAI and Planning Agent capabilities point toward a more interactive, proactive and intelligent FP&A experience.

For finance leaders, the opportunity is clear. AI in EPM should not remain centralized, technical or experimental. It should become a governed finance capability supported by security, explainability, user ownership and adoption planning.

NexInfo helps organizations adopt Oracle Cloud EPM IPM capabilities with ISO-certified delivery governance, AI-enabled transformation experience and deep Oracle EPM expertise.

FAQ

What is IPM in Oracle Cloud EPM?

IPM, or Intelligent Performance Management, is Oracle Cloud EPM’s AI and machine learning capability for finance. It supports areas such as Insights, AutoPredict, Advanced Predictions, Predictive Cash Forecasting, narrative generation and AI agents.

What is the IPM Manage role?

IPM Manage is a granular role released in 25.10 for Enterprise customers. It allows selected finance users to configure and manage IPM jobs such as Insights, Auto Predict and Advanced Predictions without relying entirely on Service Administrators.

Who can use the IPM Manage role?

The role can be assigned to individual users or in bulk to user groups, such as an FP&A or functional group. Every assigned user can create and manage their own jobs.

Does IPM Manage bypass EPM security?

No. Oracle states that IPM Manage honors existing EPM security. Users can only create and run jobs for dimension members they are authorized to access.

What IPM jobs can users manage with IPM Manage?

The role supports configuration of IPM jobs for Insights, Auto Predict and Advanced Predictions.

What is Feature Importance in Advanced Predictions?

Feature Importance helps users understand which business drivers have the strongest impact on forecasts. It compares driver contribution and supports explainability, transparency and auditability.

What is Intelligent Feature Engineering?

Intelligent Feature Engineering prepares data for machine learning by transforming existing features or creating new ones. Oracle identifies time-based features, lag effects, autocorrelation and rolling mean or rolling median transformations.

What is Intelligent Feature Selection?

Intelligent Feature Selection chooses the most relevant drivers for prediction and removes noisy or low-impact variables. Oracle identifies methods such as Random Forest Regressor, correlation coefficients, Lasso Regression, Elastic Net, Ridge Regression and Auto Mode.

What is Intelligent Data Selection?

Intelligent Data Selection uses only series with valid actuals in the driver slice for prediction. Series with missing actuals are automatically ignored to help prevent prediction job failures.

What is Dynamic Calc in Advanced Predictions?

Dynamic Calc in Advanced Predictions allows predictions at parent-level or dynamic calc members without depending on detailed leaf-level data. This helps when granular data is sparse or unsuitable for prediction.

What are In-Context Insights?

In-Context Insights surface insights directly on forms and dashboards, making AI insights available to a wider user base and helping accelerate AI adoption.

What is Periodic Movement Variance Insight?

Periodic Movement Variance Insight monitors a specific Account member, KPI or metric and triggers an insight when the recent period movement is materially significant compared to a prior-period benchmark.

What is Root Cause Analysis in IPM?

Root Cause Analysis is designed to generate a structured breakdown of contributors, potential correlations and causal factors for detected insights such as anomalies or forecast variances.

What is the Planning Agent?

Planning Agent is an AI-powered agent embedded into EPM workflows to accelerate planning, analysis and decision-making for FP&A users. It is designed to move users beyond static reports and manual analysis toward proactive, on-demand insights.

How can NexInfo help with Oracle EPM IPM?

NexInfo can help with IPM readiness, IPM Manage role governance, Insights configuration, AutoPredict and Advanced Predictions setup, forecast explainability, root cause analysis readiness, Planning Agent roadmap planning, security review, training and managed support.

Oracle Cloud EPM IPM Update: Moving Finance AI from Centralized Control to Governed Self-Service