In Part 1, we explored how Prompt Columns in Microsoft Dataverse bring generative AI directly into business data, helping organizations automatically summarize, classify, and extract insights from records inside their existing workflows.
But as AI adoption grows, businesses need more than intelligent automation, they need control.
Running AI on every record can quickly become inefficient, expensive, and disruptive. To address this, Microsoft has introduced an enhancement to Prompt Columns: filter-based execution
Together, this feature ensures AI only runs when specific business conditions are met.
The Problem It Solves
Without execution controls, Prompt Columns can trigger AI for every record update—even when the data does not require AI-generated insights.
This often creates three major challenges:
- Unnecessary AI Consumption
AI may process blank, incomplete, or low-priority records, generating little business value. - Higher Copilot Credit Usage
Every AI execution consumes Copilot credits. Running prompts without filters can quickly increase operational costs. - Workflow Delays
Real-time prompt execution across large volumes of records can slow down user actions and affect overall system performance.
With filters, organizations can eliminate these inefficiencies while maximizing AI impact.
Real-World Scenario: Smarter Customer Support Case Summaries
Consider a customer support team handling thousands of service tickets every day.
The organization uses a Prompt Column to automatically generate AI-powered case summaries for support agents. Initially, AI was running for every ticket, including simple requests like password resets or billing questions.
This resulted in unnecessary AI usage and increased costs, while many summaries provided little practical value... Read More
