What challenges did Public Service Commission of Canada face before AI implementation?

Finance experts were repeatedly interrupted by routine travel-policy questions. The process caused delays, costly last-minute bookings, and lost strategic work. The PSC needed to modernize without increasing headcount.

What was the impact of the AI solution?

Reduced travel-mailbox emails by 90%. Reduced complex query times from 5 days to 30 seconds. Created approximately $120,000 CAD per year in salary cost avoidance. Enabled earlier bookings that can save hundreds or thousands of dollars per trip.

Source

MicrosoftProductionEvidence: Medium65/100

Public Service Commission of Canada built a Copilot agent for travel policy Q&A

Use case typeRegulatory inquiry automationUpdated Aug 4, 2026

The Public Service Commission of Canada modernized its travel-approval process to reduce interruptions from routine policy questions and improve self-service for employees. The organization built a secure in-house AI agent using Microsoft 365 Copilot Chat and Copilot Studio with policy content from SharePoint.

Organization: Public Service Commission of Canada
Industry: Public Sector
Location: Canada
Published: August 2026

Reported outcomes

Travel-mailbox emails: −26–2,025%

Other quantified impact

Salary cost avoidance: Approximately 120,000 CAD/year

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Travel-mailbox emails: 26-2,025% decrease

Microsoft Customer StoryAug 4, 2026Customer storyExplicit claimMedium evidence strength

This shift is clearly demonstrated by a 90% reduction in emails to the travel mailbox since the introduction of the agent in fiscal year 2025-26.

Normalized claim

Complex query time: 10-30%

Microsoft Customer StoryAug 4, 2026Customer storyInferred claimMedium evidence strength

Previously, clients used to wait days—or, sometimes, weeks—for answers to multi-step, complex queries. Now, with the agent, travelers receive a compliant answer with citations within 10 to 30 seconds.

Normalized claim

Salary cost avoidance: 120,000 CAD/year

Microsoft Customer StoryAug 4, 2026Customer storyExplicit claimMedium evidence strength

By redirecting time to value-adding work, the department has achieved a significant cost avoidance—including the salary of an extra FTE, which would have cost approximately $120,000 CAD per year.

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Public Service Commission of Canada
Provider: Microsoft
Maturity: Production
Linked source: Microsoft Customer Story

Deployed the solution with human oversight, citations, and guardrails for out-of-scope routing

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1Regulatory inquiry automation
2Customer service automation
3Workflow automation

Built a secure in-house AI agent with Microsoft 365 Copilot Chat and Copilot Studio.
Connected the agent to SharePoint-hosted policies, SOPs, and travel documents.
Deployed the solution with human oversight, citations, and guardrails for out-of-scope routing.
Used a phased pilot to validate agent behavior before broader rollout.

Technologies

Microsoft 365 Copilot Chat Copilot Studio SharePoint

Reduced travel-mailbox emails by 90%.
Reduced complex query times from 5 days to 30 seconds.
Created approximately $120,000 CAD per year in salary cost avoidance.
Enabled earlier bookings that can save hundreds or thousands of dollars per trip.

Architecture

PSC built an internal travel-policy AI agent using Microsoft 365 Copilot Chat and Copilot Studio, grounded in SharePoint policy documents, with bilingual responses, citations, human oversight, and routing of out-of-scope questions to subject matter experts.

Sources & evidence1

Evidence: Medium65/100Evidence strength