MicrosoftI’ve deployed the Knowledge Agent across several SP sites last year, and overall, it’s been a strong step forward for AI adoption within the platform. After using it in production environments, there are a few improvements worth considering, particularly around governance, security, and content integrity.
Refine Visibility of “Improve this site” Actions
The “Improve this site” and related AI-driven management actions should be more tightly aligned with SharePoint’s governance model. In most enterprise environments, structural site improvements and AI-generated recommendations that affect content organization are typically the responsibility of site leadership.
A more controlled approach would be:
Default visibility limited to Site Owners and Designers
Optional enablement for Editors (configurable by Owners)
No visibility for Contributors and Viewers
Restricting these controls would reduce the risk of unintended structural changes and reinforce clear ownership of site quality and lifecycle management.
Gate AI-Driven “Create Page”
AI-generated prompts to create new pages are powerful but they also influence information architecture. In mature environments with defined content governance, new page creation tied to AI insights should ideally be reviewed or initiated by Site Owners. However, based on my experience, when such features are less visible to regular users, they tend to drive less active site management and support stronger governance.
…would strike a balance between empowerment and control.
