Copilot Studio Launched #315

Task completion grader in Copilot Studio

Evaluate whether agents complete their intended tasks

The task completion grader in Copilot Studio lets agent builders evaluate whether their agents accomplish intended tasks. It reached general availability on October 15, 2026.

Key points

  • Evaluation goes beyond checking whether a response is relevant or high-quality.
  • The grader helps identify functional failures.
  • Results can help builders improve agent reliability.

What to do

No action is required. Tell agent builders that the task completion grader evaluates task success beyond response quality.

Timeline

  1. 15 October 2026 General availability (confirmed) Task completion grader reached general availability.
  2. 29 October 2026 Published
Speaker notes

A relevant or high-quality response does not by itself show that an agent accomplished its intended task. The task completion grader in Copilot Studio addresses that distinction. It helps builders identify functional failures and gain confidence that agents deliver expected business outcomes. Microsoft states that the feature reached general availability on October 15, 2026. No action is required, and the source does not specify licensing requirements or admin controls.

Sources and links

Area, status, change type, licence and flags are indicative. Always open the source before you act on a card.