Currently listed with a status of In development and targeted for General Availability in November 2026, the update is scheduled for Web platforms across Worldwide (Standard Multi-Tenant), GCC, and GCC High cloud instances. While roadmap timelines represent Microsoft's current rollout plans rather than ironclad guarantees, the functional shift is clear: makers will soon be able to store longer, more detailed AI-generated insights directly inside their Dataverse table columns.
The Single-Line Bottleneck in Dataverse AI
To understand why this relatively compact schema update matters in the real world, it helps to look at how Dataverse prompt columns operate today. According to Microsoft Learn documentation, a prompt column is an AI-powered data type in Microsoft Dataverse that lets makers define natural language prompts tied to other columns within the same table. When a record is created or updated, the AI model processes the referenced input columns and stores the generated response persistently in the prompt column for use across model-driven apps, workflows, views, and reports.
Historically, however, third-party Dynamics 365 implementation specialists such as Inogic have noted that prompt columns have been created as single-line text fields, lacking multiline format support and forcing makers to stretch column widths on forms just to make longer AI outputs readable. That constraint worked fine if your prompt was merely classifying a support ticket as "Urgent" or extracting a sentiment score. It became far less practical when asking the AI to draft a customer reply, summarize a complex dataset, or generate multi-point sales guidance.
By enabling prompt columns to store multiple lines of text as output, Microsoft is aligning the storage container with the actual generative capabilities of the feature.
How Prompt Columns Work Under the Hood
While the November 2026 roadmap item expands the shape of the stored output, Microsoft Learn’s current technical documentation outlines the strict operational guardrails, prerequisites, and architecture that govern prompt columns today:
- Table and Input Limits: Makers can create up to five prompt columns per table. A prompt can reference multiple input columns from the table, but it cannot use formula columns, file columns, image data types, or another prompt column as inputs—if selected, those values are ignored. Every prompt definition also requires a default text input variable (
PromptColumnRecordIdby default) applied as a filter on the data source's primary column. - Asynchronous Execution: To prevent AI processing from bogging down real-time database transactions, prompt columns use asynchronous computation. When a record is created or when one of the specific input columns referenced by the prompt is modified, the prompt runs in the background. If unrelated columns on the record are edited, the prompt does not run.
- No Automatic Backfills or On-Demand Runs: Existing records in a table are not processed by default when you add or change a prompt column, and on-demand execution without a record update is not currently supported. Updating a referenced input column is required to trigger analysis on an existing row.
- Filter-Based Execution and Credits: Every prompt execution consumes credits—specifically AI Builder credits (which Microsoft notes are retiring soon) and Copilot Credits. While testing a prompt inside the prompt designer against sample records is free, live executions are billed. Makers can configure filter conditions on the prompt column so that AI generation only triggers when specific business criteria are met, avoiding wasted credits on irrelevant rows. Consumption can be audited via the Power Automate Automation center under AI Builder activity.
What We Know—and What Remains Unspecified
As a veteran observer of Microsoft roadmap dispatches, it is important to separate confirmed database mechanics from open questions. Microsoft’s roadmap entry confirms that prompt columns will store multiple lines of text to support longer, detailed insights, but it does not yet specify the exact character ceiling or underlying Dataverse attribute type that will back the updated prompt columns.
For context, Microsoft Learn’s general Dataverse column documentation defines three standard text column capacities:
| Dataverse Column Type | Default Max Length | Absolute Maximum Characters | Presentation Format |
|---|---|---|---|
| Text | 100 | 4,000 | Single-line textbox |
| Text Area | 100 | 4,000 | Multi-line textbox (for smaller text blocks) |
Multiline Text (MemoType) | 150 | 1,048,576 | Multi-line textbox (for large text blocks) |
Whether the November 2026 update maps prompt columns to the 4,000-character Text Area specification, the 1,048,576-character Multiline Text (MemoType) specification, or a custom prompt-specific limit is not established in the roadmap notice. Similarly, Microsoft has not yet detailed whether existing single-line prompt columns will upgrade automatically or require maker reconfiguration to take advantage of multiline rendering on model-driven forms.
Practical Preparation for Power Platform Admins
If your organization plans to deploy or expand Dataverse prompt columns ahead of the planned November 2026 rollout, Microsoft Learn documentation provides clear configuration and troubleshooting procedures you can verify right now:
- Verify Environment Prerequisites: Ensure you have appropriate AI Builder or Copilot licensing and that Copilot and AI Prompts features are turned On in your Power Platform environment settings. Additionally, the Block unmanaged customizations environment setting must be disabled when creating or modifying prompt columns, or makers will hit an
msdyn_AIConfigurationpermission error. - Configure the Prompt Column Properly: In Power Apps (Tables > select table > Columns > New column), set the Data type to Prompt, clear the Allow form fill assistance checkbox, and select +Add new prompt to build and test your instructions against existing table records.
- Monitor Execution with System Status Columns: Whenever you create a prompt column, Dataverse automatically provisions two companion tracking columns:
(columnName)_PromptColumnStatusand(columnName)_PromptColumnDetails. Adding these to your table views allows administrators to see whether a generation isNotStarted(0),InProgress(1),Completed(2), skipped due to unmet filter conditions (2000), skipped because referenced input columns did not change (2001), orFailed(3)—which commonly occurs if a user lacks read permissions on a referenced input column or if the tenant runs out of Copilot entitlements.
Giving AI prompt columns room to breathe across multiple lines is a sensible, overdue quality-of-life improvement for Dataverse builders. Once the capability reaches general availability in Worldwide, GCC, and GCC High environments, makers should finally be able to store structured summaries and multi-paragraph analyses directly inside their records without fighting single-line UI constraints.
References
- Power Apps: Prompt Columns - multiple lines of text output Microsoft 365 Roadmap · 2026-10-15T22:47:30.226767Z
- Create and edit columns in Dataverse using Power Apps - Power Apps | Microsoft Learn learn.microsoft.com
- Prompt columns in Microsoft Dataverse - Power Apps | Microsoft Learn learn.microsoft.com