Concept: Planning Agent

The Planning agent is an intelligent, conversational AI assistant that integrates directly into Adaptive Planning using Ask Workday for Adaptive Planning (Ask Workday). Acting as an on-demand digital analyst, it empowers FP&A professionals, CFOs, and operational managers to interact with complex multi-dimensional models using natural language.

Using agents in a sandbox Adaptive Planning instance doesn’t consume flex credits.

The AI feature enables you to:

Deliver immediate, contextual insights for ad-hoc and variance analysis, enabling cross-functional teams to quickly:
Surface trends
Identify root causes
Drive more informed decisions
Reduce risk of mis-analysis by surfacing data highlights and anomalies that you might otherwise miss manually.
Empower business units to self-serve analyses without FP&A or admin bottlenecks.
Streamline data storytelling with instant visualizations and narrative summaries.
Use the Ask Workday prompt to:
Enter the @ symbol to trigger a pop-up menu of element types such as accounts, dimensions, or versions.
Enter partial names, abbreviations, or minor typos and still find the correct dimension such as level or account.

The Analyst role of the Planning Agent provides these agent skills:

For Adaptive Planning standalone instances, admins must assign the

Access Ask Workday for Adaptive Planning

permission to users for accessing the Data Exploration skill. This parent permission includes these child permissions:

Access Planning Agent
Data Exploration Skill for Planning Agent

With the

Access Ask Workday for Adaptive Planning

parent permission assigned, users can view the

Ask Workday

icon on all application pages including standard, model, and cube sheets.

Workday collects operational telemetry and agent execution traces to:

Monitor system behavior and performance.
Improve the product experience, guide manual improvements, and ensure system reliability over time.

Workday doesn't use the data to retrain or fine-tune the pre-trained foundation model. You can disable the collection of this trace data at any time by selecting

Administration

General Setup

Disallow Data Contribution

.

The Planning Agent:

Queries the live data from your Adaptive Planning model ensuring that analysis always reflects the latest data entries, imports, and formula adjustments.
Strictly respects existing access rules and permissions. Users can only retrieve or analyze data they are explicitly authorized to view in the system, ensuring data governance.
Can pivot multiple dimensions simultaneously and intelligently structure the responses. If the pivot becomes too complex for readable visualization, it advises on granularity.
Example: "Trend units by customer by month for FY 2024".
Includes an integrated search tool that you can trigger by the @ symbol. The search enables users to browse and select specific model elements directly within the prompt and adds precision to queries.

Planning Agent:

Treats each prompt as a new query and doesn't yet support multi-turn follow-up questions.
Uses Generative Artificial Intelligence (GenAI), an evolving technology. Although built and tested for high accuracy, it can occasionally misinterpret questions or present unexpected results. Always review the output including numbers, commentary, and visualizations. Before acting on the results, verify them against the source of truth.
Only understands questions and responds in US English.
Reloads when you navigate to a different location or the session ends.
This means we erase the conversation and all responses.
Provides suggested questions based on your last refresh.
Example: If a report displays Years, which you expand to show Months, the agent continues to make suggestions based on Years, rather than Months, until you refresh.
Treats each prompt from you as a new prompt. You can’t yet build upon questions and answers.
Doesn’t provide a complete list of suggested questions.
Example: Suggested questions don’t include attributes, but you can still ask questions and get answers about attributes when you ask open-ended questions.