Skip to main content
Aqlyst Technologies — home

AI Agents • Microsoft Copilot

Enterprise Knowledge Agent

A governed Copilot Studio experience that answers policy and procedure questions from approved sources, cites where each answer came from, and hands routine requests straight into the systems that process them.

Client
National research organisation
Industry
Research and healthcare
Start a conversation

Note. Portfolio concepts shown here are anonymised and representative of the kinds of intelligent business systems Aqlyst delivers. Client names, screenshots, and measured results are published only with written approval.

Challenge

The situation

  • Knowledge spread across disconnected systems

    Policies, procedures, and guidance lived across SharePoint sites, shared drives, and long email threads. Staff could rarely tell which version was current, so most questions ended up routed to a small group of specialists.

  • A support queue absorbing routine questions

    A significant share of incoming requests were repeat questions with a documented answer, which delayed the cases that genuinely needed expert attention.

Approach

How we worked

  1. Scenario selection before toolingWe started from the questions people actually asked, grouped them by system of record, and selected a first scope that was valuable, bounded, and answerable from content with a named owner.
  2. Grounding on approved content onlyGrounding was limited to reviewed source locations. Content without an owner or a review date was deliberately excluded rather than included and hoped for.
  3. Permission inheritance rather than a service accountThe agent operates under the signed-in user's existing permissions, so it cannot surface material that person could not already open.
  4. Evaluation set built before launchA test set of real questions with expected answers was agreed with subject-matter experts and re-run after every meaningful change to grounding or topics.

Solution architecture

How the system fits together

Layered agent architecture: an experience layer in Microsoft 365, an orchestration layer in Copilot Studio, actions through Power Automate, and grounding limited to reviewed sources.

  1. Experience

    • Microsoft Teams
    • Microsoft 365 Copilot
    • Web chat

  2. Agent

    • Copilot Studio topics
    • Orchestration
    • Escalation rules

  3. Actions

    • Power Automate
    • Dataverse
    • Service request system

  4. Knowledge

    • Approved SharePoint sites
    • Policy library
    • Azure AI Search

  5. Governance

    • Microsoft Entra ID
    • Evaluation set
    • Usage monitoring

Delivered

What was built

  • Grounded question answering with citationsAnswers link back to the source document and section, so the reader can verify the response rather than trust it.
  • Business actions inside the conversationPower Automate actions let the agent raise a request, update a record, or start an approval without the user leaving the chat.
  • Escalation to a humanLow-confidence and out-of-scope questions route to the right team with the conversation context attached.
  • Governance and monitoring packDocumented grounding scope, action inventory, permission model, review cadence, and a dashboard of unanswered questions.

Outcomes

What changed

Figures are published only once the client has approved them. Where a value reads “Illustrative”, the improvement is real but the number has not been cleared for publication.

Time to a trusted answer
Not yet approved for publication
Routine questions handled in-channel
Not yet approved for publication
Answers with a verifiable source citation
All in-scope responses

Technology stack

  • Copilot Studio
  • Azure AI
  • Power Automate
  • Microsoft 365
  • SharePoint

Related solutions

  • AI Agents & Copilots

    Governed agents that answer questions from trusted content and take real business actions, built on Copilot Studio, Microsoft 365 Copilot, and Azure AI.

    Explore AI Agents & Copilots

  • Azure Cloud & AI

    Cloud application architecture, integration, identity, and AI services designed for security, observability, and predictable operating cost.

    Explore Azure Cloud & AI

All solutions
Next case studyLifecycle Operations PlatformRead next

Let's talk

Working on something similar?

Bring the problem as it actually is, including the parts that are messy. That is usually the fastest route to a useful first conversation.

Start a conversationView all work

Prefer email? info@aqlyst.ai