HR policy questions answered in Teams — without emailing HR
A Copilot Studio agent connected to SharePoint that answers policy questions in plain English, always cites the current authorised version, and escalates sensitive queries to HR.
See it working
Demo video — coming soon
Policy documents exist. Nobody can find the right one.
Most organisations have a SharePoint intranet with hundreds of policy documents. Employees know the policies exist somewhere — but finding the right document, navigating to the correct section, and confirming it's the current version takes time they don't spend. Instead, they email HR. HR answers the same questions dozens of times a week.
For a 200-person professional services firm, the HR team was fielding roughly 40 policy-related queries per week. Each query took the HR team an average of 15–20 minutes to respond to, including finding the document, reading the relevant section, and writing a reply. That's 10–13 hours of HR time per week on questions the policy documents already answer.
A Copilot Studio agent in Teams, backed by SharePoint
The agent lives in Microsoft Teams — where employees already work. It knows nothing except what's in the authorised SharePoint policy library, and it always tells you exactly which document it's reading from.
Employee asks a policy question in Teams
"What's our remote working policy?" or "How much notice do I need for annual leave?" — questions asked in plain English, not search syntax.
Agent queries the SharePoint knowledge source
Copilot Studio retrieves semantically relevant content from the authorised policy library. Only documents published to the live library are ever used — draft and archived versions are excluded.
Answer returned with citation
The agent responds with a plain-English answer and includes the document title, section, and a direct link to the SharePoint file. Employees can verify the source in one click.
Sensitive queries escalated to HR
Questions about grievances, disciplinary processes, sickness absence, or any topic flagged as sensitive are not answered by the agent. It acknowledges the query and routes to a named HR contact via Power Automate.
Unknown queries flagged for HR review
If the agent can't find a relevant policy it tells the employee honestly and logs the query for the HR team to review.
Results at Week 4
Baseline measured in the 4 weeks before deployment. Post-deployment measurement at week 4.
| Metric | Score | What it measures |
|---|---|---|
| Policy emails to HR | 73% fewer | Week 4 vs pre-deployment baseline |
| Avg time to answer | 3 min | Down from 20+ minutes search + wait |
| Answers citing live documents | 100% | Always the current authorised version |
| HR hours saved | ~9 hrs/week | Redirected from policy queries to complex HR work |
What it doesn't handle well
The agent deliberately doesn't answer questions about individual employment terms, ongoing grievance or disciplinary matters, or anything requiring HR judgment rather than policy lookup. Two further edge cases surfaced in use: questions about policies for the firm's recently acquired subsidiary (whose documents weren't yet in the SharePoint library — fixed by HR in week 2) and ambiguous questions that matched multiple conflicting policies from different regions. The latter is rare but the agent now flags a warning in those cases rather than picking one.
What it's built on
Common questions
How does the agent know which version of a policy is current?
It reads directly from the SharePoint library at query time — not from a cached copy. The HR team manages version control in SharePoint as they always have.
What happens to documents with restricted access?
The agent inherits SharePoint permissions. If a document is restricted to HR only, the agent won't serve it to employees who don't have access.
Can it handle questions across multiple policy documents at once?
Yes. If a question touches multiple policies, the agent synthesises across relevant documents and cites each one separately.
If you're on Microsoft 365, a Copilot Studio policy agent can be live in 2 weeks.
Book a 25-minute call. We'll map your policy library and show you what the agent would look like in your Teams environment.