Microsoft Copilot Studio UX - Build 26.06.14-17

This article describes the fixes and improvements to existing functionality that are included in this build.

Visit the Microsoft Copilot Studio Blog to stay up-to-date on the latest news!

Version 26.06.14-17 updates

New features

• Workflow Analytics now includes new charts for flow run trends and actions over time and enables custom metrics drilldown by default.
• Agents using cascade mode can configure Text-to-Speech voice settings per language, with sensible default voices applied when enabling voice, adding languages, and switching to cascade mode.
• Workflow evaluation results include a flow execution trace in the results details, and a run actions menu to rename or delete completed runs and stop or cancel in-progress runs.
• Makers can quickly repeat studies with an Evaluate again action that opens a new evaluation pre-filled with the prior workflow runs and test methods.
• Evaluation views provide clearer test-method insights with split columns showing methods and scores, along with general response quality metrics and improved explanations in test run details.
• Analytics for enhanced task completion agents (Preview) adds new cards for sentiment, connected agents, and plugin actions.

Fixes and improvements

• Accessibility and usability improvements: stable and descriptive screen reader labels, better keyboard focus handling, accessible filter group labels, clearer adaptive card announcements, Voice Access to sensitive data tooltips, headings that wrap on small screens, variable picker menus that remain fully visible, info popovers that stay within the viewport at 200% zoom, and focusable links in dialogs.
• Maker Evaluation and results UX: non-overlapping grid headers that remain visible while scrolling, updated layouts and wording, clearer score badge behavior, a new Status column, improved loading indicators, disabled navigation for unprocessed items during live runs, more readable durations, improved spacing and scrolling in results modals, and correct “Show more” behavior for many graders.
• Workflow evaluation operations: immediate feedback with a loading spinner on Evaluate, consistent workflow run names, the ability to export workflow evaluation run results, updated background styling, and removal of unintended sorting when clicking info buttons.
• Analytics and sessions: Analytics pages load even for unpublished agents, session transcript downloads are more reliable, the main Analytics page uses a See details button that leads to the Sessions page for downloads, adaptive cards opened from transcript pop-ups are read-only, Workflow Analytics shows loading placeholders and 7, 14, and 30 day date presets, and wording is clarified for Copilot credits billing and answer quality labels.
• Agents, tools, and publishing: the Agent Status page title no longer shows “(preview)”, only published Fabric data agents are listed, tenant-specific pages respect the tenant in the URL, custom connectors and custom MCP tools are no longer incorrectly disabled while prebuilt connectors still follow policy, the Tools experience highlights blocked connectors (with indicators in-flow), deep links open classic flows in the correct designer, Real-time Voice is shown as not supported for the Teams Phone channel (with certain auth options locked after configuration), the publish confirmation dialog is restored, and license upgrade banners no longer flash for eligible users.
• Performance and reliability: topic loading and editing is significantly faster for large agents, the AgentOps page reliably reflects the latest results with workflows in inventory, and continued improvements to performance and reliability.

Security

• No changes in this release.

Return to the Microsoft Copilot Studio Release Notes.