About Decision Intelligence
Decision Intelligence (DI) enables analysts to use AI to investigate data, explore what-if scenarios, and submit changes to Adaptive Planning through existing security and approval chains. DI uses your existing chart of accounts, consolidation hierarchy, organizational structure, and security context. You can analyze financial data without exporting information, setting up external data connectors, or engineering prompts.
You access DI from the Global Navigation menu in Adaptive Planning. Prompt DI in plain language to analyze and model against your data. Every computation runs in an isolated environment with no access to external networks or other tenants' data. You can use DI to:
- Investigate variances and receive structured breakdowns by dimension, driver ranking, and contribution analysis.
- Analyze root causes using mix-shift analysis, correlation analysis, and statistical hypothesis testing to decompose performance shifts across dimensions.
- Model and compare scenarios by building what-if cases as assumption deltas on a baseline.
- Run Monte Carlo simulations to model probability-weighted outcome ranges across key assumptions to quantify risk and confidence intervals in your forecasts.
- Reforecast from drivers by projecting financial outcomes from operational inputs, such as pipeline, headcount, and conversion rates, using regression analysis and trend modeling.
- Apply advanced analytics, including linear regression, time series decomposition, cohort analysis, and Pareto analysis, using natural language.
- Generate narrative summaries grounded in your analysis, including variance explanations, scenario comparisons, and quarterly performance summaries.
Administrator Role
As an administrator in DI, you can configure and manage features for your tenant, including:
- Organization Memory, which enables you to provide guidelines and conventions that every chat should follow.
- Manage data source connections. You can connect Snowflake to DI for your tenant so that all users can connect to it without additional configuration.
- Design systems, which is only available through Labs at this time. You can use the Design Systems feature to create new design systems that all DI users can then select from their dashboard toolbars. You can also set the default design system for the tenant.
To set up the administrator role:
- Click theGlobal Navigationmenu in Adaptive Planning, and then clickAdministration.
- ClickPermission Sets, and then clickEditfor theAdministrative Permissionset.
- Select theManage Decision Intelligencecheck box under theAccess Decision Intelligencepermission.
Data Sources
You can bring your data into DI by:
- Uploading the following file types directly in a prompt by clickingPlusand thenAttach:
- Microsoft Excel (.xls, .xlsm, .xlsx)
- CSV (.csv)
- PDF (.pdf)
- Microsoft Word (.doc, .docx)
- Markdown (.md)
- Text (.txt)
- Adding Matrix Reports from Adaptive Planning to a project by clicking , and then selecting the reports you want to add. DI works with both the report’s definition and data, enabling you to expand report details and rerun reports with different parameters.
- Using your existing Adaptive Planning standard sheets, modeled sheets, and cube sheets, including any dimensional intersection. When DI pulls data from Adaptive Planning, it also fetches the relevant structures.
- Connecting to Snowflake so that DI can pull your Snowflake warehouse data into an analysis alongside your Adaptive Planning data or uploaded files. See Connect Snowflake to Decision Intelligence.
When you combine multiple data sources in a single project, such as a Salesforce pipeline export and your Adaptive Planning forecast, DI joins them automatically using shared dimensions like time period, region, or cost center. You can run cross-source analysis in a single conversation. Example:
Using the pipeline data I uploaded and the Q4 plan from Adaptive, show me where the revenue forecast is at risk.
When you bring data into DI, it reads your files or Adaptive Planning data and builds a semantic model of your data's structure that ensures that every query DI runs is valid and traceable. You can:
- Review the model in the model panel to see dimensions, measures, formulas, and how datasets connect. There are 2 ways to view the model:
- Click the project name header and selectView model.
- Click the folder icon in the upper-left corner of the canvas and click the model.
- Correct the model through conversation. Example: If DI misclassifies a column, you can provide a correction. DI then updates the model and saves the correction to the project.
- Define formulas, which are computed measures that derive from existing data. Example: Gross Margin = Revenue - COGS. DI stores formulas in the model and applies them to all future queries.
Calculation Traceability
Every computed number in DI traces back to source data. You can ask DI to explain any result to view the full calculation chain, including:
- The data DI used.
- The formula DI applied.
- How DI derived intermediate results.
Example:
Show me exactly which rows you used for that number
.Calculation traceability works across the full analysis chain. If a variance breakdown feeds into a scenario, DI can trace the scenario output back through the variance calculation to the original data.
If a number looks wrong, you can ask DI to show its reasoning and correct it if needed.
Projects
Starting a new conversation in DI creates a project, which is a persistent workspace that keeps your data, semantic model, conversation history, and analysis artifacts together. You can open an existing project and conversation from the home page or the navigation menu to continue your analysis where you left off.
You can view a project's full inventory, including its sources, artifacts, and semantic model, by clicking the folder icon at the top of the canvas. From the inventory, you can open and close artifacts, including dashboards created throughout your conversation, as tabs in the canvas. When you update a dashboard, DI gives it a new version. DI displays the most recent version of a dashboard when you open it, but you can access previous versions from the version list when the dashboard is open. You can rename an artifact by double-clicking its tab or clicking the
More Actions
menu on the tab. You can also pin an artifact from the More Actions
menu to keep it always on top.Playbooks
A playbook is reusable analysis you can run in one click, rather than writing a prompt from scratch. You can use Workday-created playbooks or create your own playbooks.
DI includes 2 Workday-created playbooks:
- Headcount Walk Analysiscompares 2 workforce versions and produces a canvas of the headcount walk. It shows the headline position, where the change is concentrated, and a filterable register of added, removed, reassigned, and reclassified positions.
- Budget vs. Actuals Variance Analysisdelivers full-P&L budget-to-actual variance analysis as a 3-tab canvas, including:
- A variance overview with a budget-to-actual bridge.
- Ranked contributors.
- Drivers with next steps.
You can open a playbook from the:
- Navigation menu.
- Chat prompt by:
- Clicking and selecting a playbook.
- Entering a forward slash (/) and selecting a playbook.
You can add custom instructions when you run a playbook to adjust the analysis, such as specifying a region, time period, or focus area. You can add instructions to a playbook:
- In theAdditional instructions (optional)field when you open the playbook from the navigation menu.
- In the chat prompt when you open a playbook using the forward slash. Example: /budget-vs-actuals for the west region.
You can create your own playbook by:
- Clicking .
- ClickingPlaybooksin the navigation menu, and then clickingCreate Playbook.
- Entering/create-playbookin the chat prompt.
A workflow helps you create your playbook by asking you questions about the data you’re looking for and how you want it presented. Provide your answers in the prompt. When your custom playbook is created, you can run it by:
- Clicking and selecting your playbook.
- Entering a forward slash (/) and entering the name of your playbook.
You can edit a custom playbook by clicking
Playbooks
in the navigation menu, clicking the More Actions
menu on your playbook, and selecting Edit
.Outputs and File Exports
You can edit DI’s analytical outputs by changing chart types, sorting and filtering data, and drilling into specific dimensions. Each change you make creates a new version of the output. You can switch between versions to compare changes and decide which one works best.
DI can generate these visualization types:
- Bar charts and grouped bar charts
- Waterfall (bridge) charts for contribution decomposition
- Line charts for trends and time series projections
- Scatter plots and correlation matrices
- Tornado charts for sensitivity rankings
- Tables with inline data
- KPI dashboards with delta indicators
- Monte Carlo probability distributions with confidence intervals
- Multi-panel interactive dashboards with filters, sliders, and drill-down controls
DI also can produce finished deliverables in the format you need, such as a board summary, a variance report, or a scenario comparison deck. Example:
Create a PowerPoint summarizing the Q3 variance analysis with the waterfall chart and my recommended reforecast.
DI generates a downloadable file containing your analysis. DI supports these file formats:
- PDF (.pdf)
- Microsoft PowerPoint (.pptx)
- Microsoft Word (.docx)
- Microsoft Excel (.xls, .xlsm, .xlsx)
- CSV (.csv)
- Markdown (.md)
- JSON (.json)
- YAML (.yaml)
You can also export the underlying data as CSV from any table.
Submit to Plan
When you're ready to act on a scenario, DI displays a structured preview of the proposed changes, including:
- Every cell that will change, organized by account, level, time period, and any custom dimensions.
- The current value and the new value for each cell, including the delta.
When you approve the changes, DI submits them to Adaptive Planning along with the full context of how the numbers were generated. DI checks the underlying plan data to determine if it has changed since your analysis began, and warns you if it has before submitting the changes to Adaptive Planning.
Memory
The Memory feature enables DI to save what it learns about you and your organization and use that information in your conversations. As a user, you can enable Personal Memory for yourself, and as an administrator, you can add Organization Memory for your tenant. To enable:
- Personal Memory, click your user profile to access theSettingsmenu, and then clickMemory. You can enable theUse Personal Memorysetting on theMemorypage. When Personal Memory is enabled, you can enter information that you want DI to remember in the chat and DI saves it for use in future conversations. Example:Remember: I always want to see variance in USD and percentage.
- Organization Memory, you must have an administrator role. AccessMemoryfrom theSettingsmenu to add guidelines and conventions for your organization’s memory. Once you’ve added guidelines in theOrganization Guidelines & Conventionsfield, they are enabled for the tenant. Example: Define when your organization’s fiscal year starts or how you define gross margin.
DI Data Contribution
Workday collects operational telemetry and agent execution traces to monitor system behavior and improve the product experience. Workday doesn’t use your data to retrain or fine-tune the foundation model.
You can disable the collection of this data:
- For Adaptive Planning standalone instances, select .
- If your Adaptive Planning instance is Workday-enabled, access theData Contribution for Workday AIreport to disable your data contribution. See Manage Workday AI Data Contributions for Universal Main Subscription Agreement Customers.
Labs
The Labs feature enables you to use features that are still in development in DI before they're released to Production. The early-access features might be changed or removed from DI at any time.
To turn on Labs, click your User Profile and then click
Settings
. In the Advanced
section, enable the Labs setting. All features available through Labs have a Labs
label next to them in DI to distinguish them from Production features.These features are included in DI when you enable Labs:
- New Canvas Features: From the new canvas toolbar you can:
- Choose from 3 DI-provided design systems: Workday, Fjord, and Monochrome.
- Edit widgets in an interactive Style panel.
- Select a widget to move to the chat to interact with it or change its display.
- Custom design systems, which enables an administrator to create new design systems that all DI users can then select from their dashboard toolbars. An administrator can also set the default design system for the tenant. AccessDesign Systemsfrom theSettingsmenu.
Limitations
Feature | Limit |
|---|---|
File upload size | 50 MB total per message (resets each message) |
Files per prompt | 10 |
Prompt length | 100,000 characters |
Supported file types | .csv, xlsx, .xls, .xlsm, .doc, .docx, .md, .pdf, .txt |
Submit to plan | DI can submit to Adaptive Scenarios, and assist you in merging the scenario into the base version. Submits to all sheet types, but can’t create or manage splits. |
What-if scenarios | Adaptive Planning limitations for what-if scenarios also apply to DI. See Reference: Adaptive Planning Limits. |
DI parses cell data, formulas, number formats, named ranges, and table structures from Excel files. Macros in .xlsm files are accepted at upload but not executed.
DI doesn’t support:
- Scheduled or recurring analyses.
- Project sharing. Projects are per-user. To share results, you must export files.
- Real-time Adaptive data sync. DI fetches Adaptive Planning data at the time you request it and doesn’t update the data if the underlying plan changes during your session. When you submit, DI catches changes made since your analysis began. It displays the current values in the approval preview, which you can approve or reject.
- File uploads beyond spreadsheets.
- Personalization. Your preferred chart types, default dimensions, and narrative style don’t carry over between projects.
- Web search.
Troubleshooting
Issue | Solution |
|---|---|
DI provides a number that seems incorrect. | Prompt DI to provide specifics on the calculation. Example: Show me exactly which rows you used for that number . I can display the reasoning for the answer and the specific data behind any result. If the semantic model needs correction, tell DI directly. |
A file upload failed. | Verify that:
|
DI is slow or unresponsive. | Complex analyses with large datasets take longer. Refresh the page if DI is unresponsive. If the issue persists, start a new prompt in the same conversation. Your project and history are preserved. |
Analysis doesn't match expectations. | Rephrase your question with more specifics, including the time period, dimensions, and measures you want. Open the model panel to verify that the semantic model is correct. If a column is misclassified, tell DI. Example: Region should be a dimension, not a measure . |
Lost work or missing history. | Project history persists across sessions. Return to the project from the projects list. If a conversation is truly missing, report the issue to Workday with the full page URL. |