Microsoft has rolled out a Copilot-enhanced topic assignment engine for Viva Glint, its employee engagement survey platform, fundamentally changing how open-ended comments are classified before they appear in reports. The feature, listed as Microsoft 365 Roadmap ID 553224, became generally available in July 2026 for worldwide standard multi-tenant cloud customers.

How the New Topic Assignment Works

Rather than adding a chatbot, Microsoft has rebuilt the underlying pipeline that maps survey comments to Viva Glint’s standard topic set. Previously, a traditional natural-language processing (NLP) engine analyzed responses. If a comment didn’t clearly fit a predefined category, it risked landing in an unclassified “N/A” bucket, effectively hiding it from topic-level reports.

Now, when an organization enables Copilot in Viva Glint, comments are routed through an AI model that also considers the originating survey question. This context allows the system to distinguish between, say, a complaint about “lack of support” from a manager versus from IT infrastructure. The result: fewer comments fall through the cracks, and themes better reflect the actual feedback submitted.

The experience remains seamless for end users. Both standard and Copilot-enhanced assignments feed into the same Topics section of the Comments report. Managers and HR analysts still review aggregate sentiment, filter by demographics, and drill into representative quotes. The upgrade is a backend swap, not a UI overhaul.

Why This Matters for Employee Listening Programs

Employee survey comments are messy. A single response might touch on management, workload, communication, and well-being. Traditional keyword-based classification often struggles with ambiguity, sarcasm, or compressed language. The N/A problem was more than an annoyance — it meant potentially critical signals were getting lost.

Microsoft’s own data underscores the stakes. In 2022, 57% of Glint survey respondents left at least one comment, with unfavorable respondents more likely to comment than favorable ones. For large organizations, thousands of comments can pile up, and managers rarely have time to read them all. Topic grouping is a necessary filter, but only if it’s accurate.

By improving alignment between comment meaning and topic labels, Copilot-enhanced assignment makes it more likely that a concern about “constant reorgs and unclear priorities” gets tagged as “Organizational Change” rather than vanishing into N/A. That helps leaders spot emerging trends sooner and act on what employees are actually raising.

What It Means for HR Teams and Managers

For HR analysts, the upgrade should reduce manual work. Instead of combing through unclassified comments to find hidden themes, they can trust that the topics dashboard offers a more complete picture. When preparing executive summaries or action plans, they’ll spend less time second-guessing the classification.

Managers who rely on Glint reports to guide team discussions will see the same interface but with potentially richer insights. A topic might surface that previously went unnoticed, allowing for more focused follow-up. However, the quality of those insights still hinges on human judgment. Microsoft explicitly warns that AI-generated topics can be inaccurate and should be validated against the underlying comments.

This is especially critical when comparing survey cycles. A spike in a topic could signal a real shift in sentiment or simply better classification. Without context, leaders might misinterpret the data. HR teams should document when they enable Copilot and educate stakeholders about the methodology change.

The Privacy and Accuracy Guardrails

Employee surveys are sensitive, and Microsoft has baked in several protections. The Copilot-enhanced pipeline operates entirely within Viva Glint’s service boundaries, using only data already available to the platform. No external sources are introduced, and raw responses aren’t used to train foundation models. One organization’s comments and topic patterns are never shared with another.

Crucially, all existing confidentiality thresholds and suppression rules remain in force. If a topic doesn’t meet the minimum respondent count (typically 10, but configurable), related comments are hidden — even if a keyword cloud might still show frequent terms. Copilot does not bypass these rules.

Permissions still dictate what Copilot can process. It only works with data the signed-in user is allowed to see. That means role design matters more than ever. Access to view comments, filter by demographics, and export data should be reviewed to prevent unintended exposure.

Next Steps for Microsoft 365 Administrators

The feature is automatically active when Copilot in Viva Glint is turned on — there’s no separate toggle. Administrators control it through Viva Feature Access Management (VFAM) in the Microsoft 365 admin center, where Copilot is enabled by default but can be restricted organization-wide or by group.

Before relying on the new topic assignments for high-stakes reporting, admins should:

  • Run a pilot on a recurring engagement or pulse survey, comparing Copilot-assisted topics with manual review of the same comments.
  • Check whether formerly N/A comments now map to meaningful categories.
  • Validate that high-volume or sensitive themes (like fairness or inclusion) are classified correctly.
  • Update internal documentation with the enablement date so trend analyses account for the methodology shift.
  • Reinforce with people managers that topic labels are a starting point, not a conclusion, and should always be cross-checked against representative comments.

Copilot-enhanced topic assignment is supported for recurring and ad hoc survey programs but not for Always-On or Employee Lifecycle programs (as of the July 2026 updates to the Copilot Comments Summary feature). Administrators should verify support for their organization’s specific program types.

Outlook

This release is more than just another AI add-on — it addresses a persistent pain point in employee listening: turning unstructured feedback into reliable signals. As organizations navigate hybrid work, return-to-office tensions, and rapid technological change, the ability to surface what employees are actually saying becomes a strategic advantage. Copilot-enhanced topic assignment raises the floor for classification accuracy, but it’s not a set-it-and-forget-it solution. The organizations that benefit most will be those that pair it with disciplined analysis, clear privacy communications, and a genuine commitment to acting on feedback.