← All project stories

Case study / 2025

Company Policy Bot

Internal employee support experience for a policy-heavy operations environment.

Copilot StudioSharePointDataversePower Automate

01 / Context

The problem

Employees needed self-service access to policy documents without creating a dependency on HR or operations teams for repeat questions.

A support-heavy internal team had policy answers spread across documents, updates, and repeated employee questions.

02 / Contribution

What I built

Built a Copilot Studio agent grounded in approved knowledge sources, with topic routing, response guardrails, and escalation paths.

The design decision

The useful path was not to create a generic chatbot, but to design a guided policy assistant that knew where its answers came from.

Scope

  • Employee self-service for policy, SOP, FAQ, training, form, and departmental knowledge
  • Access through Microsoft Teams and SharePoint-facing channels
  • Grounded answers from approved SharePoint document libraries

03 / Structure

How it works

Designed as an enterprise knowledge assistant that lets employees ask natural-language questions across trusted organizational documents.

  • Copilot Studio handles conversational orchestration, topic handling, context, generative answers, and safety controls
  • SharePoint acts as the source of truth for indexed policy and departmental documents
  • Microsoft Entra ID and SharePoint permissions keep access aligned with organizational roles

Through the workflow

  1. Employee opens the assistant in Teams or SharePoint
  2. Copilot Studio interprets the query and retrieves relevant SharePoint knowledge
  3. The response is generated from authoritative content and returned in the conversation

04 / Operation

Controls and upkeep

  • Single sign-on with organizational accounts
  • SharePoint permission trimming
  • Development, UAT, and production environment path

Keeping it useful

The knowledge lifecycle stays manageable because updated SharePoint documents can be re-indexed and reused by the assistant.

05 / Result

The outcome

Provided a safer first stop for knowledge-heavy policy support using approved organizational content.

Employees gained a safer first stop for policy questions while the support team kept control over source content and escalation.

Discuss a similar project
Inside the project