Copilot Studio February 2026 adds Claude models and prompt controls
Diesen Beitrag auf Deutsch lesen
The February 2026 Copilot Studio update adds Claude model choices, prompt moderation sensitivity, consolidated prompt editing, and richer ticket-based Graph connector responses.
Original article from March 7, 2026: What’s New in Copilot Studio – February 2026 Update, by Rachit Garg, on Brewing Thought. Read the original
Source may be outdated. The original article is more than 60 days old. Dates and availability refer to the state of the source.
This is our own summary, not a republication or full translation.
TL;DR
Copilot Studio's February 2026 update improves ticket-based Microsoft 365 Graph connector responses for ServiceNow and Azure DevOps data. It adds Claude Sonnet 4.5 beta for Computer Use agents, Claude Opus 4.6 and Sonnet 4.5 in prompt builder, per-prompt moderation sensitivity, and in-agent prompt editing.
Why it matters
- Makers: Test Claude model choices and edit prompts, inputs, knowledge sources, and tests in the agent interface to refine agent behaviour.
- Admins/IT: Set moderation sensitivity per prompt where sensitive-content filtering needs to match the scenario.
- Makers: Use ticket-based Graph connector responses to present clearer summaries and actions from ServiceNow tickets or Azure DevOps work items.
Frequently asked questions
How do ticket-based Microsoft 365 Graph connectors improve Copilot Studio agent responses?
They let agents return clearer summaries and actionable insights from sources such as ServiceNow tickets and Azure DevOps work items. This reduces reliance on raw records.
Which Claude models can makers select in Copilot Studio prompt builder?
Prompt builder supports Claude Opus 4.6 and Claude Sonnet 4.5. Claude Sonnet 4.5 is also available in beta for Computer Use agents.
Can Copilot Studio set content moderation sensitivity for each prompt?
Yes, moderation sensitivity can be configured per prompt. The setting controls filtering for sensitive content including hate speech, violence, and self-harm.