AI, in your work.

On November 6, 2026, GitHub announced that the Agent adoption chart in the Copilot impact dashboard now includes Copilot Chat and the VS Code Agents window. The breakdown covers nine experiences. According to the official announcement, aggregate organization and enterprise 28-day reports also add `chat` and `vscode_agent` entries under `copilot_feature_engagement.totals_by_feature`.

The update broadens the activity shown; it does not change adoption phases or cohort sizes. Chat-first users also replaces the Passive users label.

Feature engagement and adoption phases measure different things

The breakdown considers a person engaged with a feature when they use it on at least two days during the 28-day period. One person can appear under several features, as GitHub explains in its September announcement. The categories represent counts that can overlap.

The cohort documentation distinguishes that usage from the activity required to qualify for a phase. Chat-first users groups active people who have not used a qualifying feature on at least two days. The API value remains `No Cohort`.

Using chat or the Agents window does not, by itself, count toward phase progression. Agent edits can count: making changes to files and conversing with an agent are different signals. The chat breakdown also includes agent mode, so people making agent edits appear under chat too.

Synthetic example: two feature counts can represent one person

This Nexo example uses fictional people and a 28-day window. It assumes that only the activity described occurs; it does not represent an organization's data or results.

Access, reports, and comparisons over time

The dashboard access guide requires the Copilot usage metrics policy to be enabled. Enterprise owners and billing managers, organization owners, and people with custom roles granting the relevant Copilot metrics viewing permission can access the data.

This breakdown is available in aggregate organization and enterprise 28-day reports; it is not added to user-level reports. The official API supplies the dates covered by each report. Those dates help identify the actual interval being compared.

Reports for periods before the change retain the original seven features. Nexo therefore recommends recording the coverage change and comparing the same features when examining historical trends: chat's absence from an older report does not establish that its usage was zero.

Nexo's interpretation: better questions about work

In the example, Person B appears under two features and remains one person. Adding the categories as if they represented unique users would inflate the total. Person A, meanwhile, uses chat regularly without qualifying for an adoption phase.

Our interpretation is that the broader coverage helps investigate which tasks people handle through conversation and where training might help. These signals alone do not demonstrate improved productivity or a financial return. Supporting those conclusions requires additional outcomes and analysis that accounts for differences between tasks, teams, and reporting periods.