What challenges did PwC face before AI implementation?

Difficulty integrating and orchestrating AI agents across broad ecosystems of applications, data, and partners with governance, security, and observability challenges.

Source

MicrosoftEvidence: Low35/100

PwC and Microsoft Azure AI Foundry Agent OS for Enterprise AI Agent Orchestration

Use case typeWorkflow automationUpdated Jun 13, 2026

PwC developed an agent OS accelerator built on Microsoft Azure AI Foundry to provide a scalable, secure orchestration layer for AI agents across heterogeneous platforms.

Organization: PwC
Industry: Professional Services
Location: United States
Published: March 2026

Planned next steps

The source says the organization plans to achieve Revenue: Up to 90% higher.
The source says the organization plans to achieve Revenue: Up to 40% higher.

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

Normalized claim

Revenue: 90% increase

PwCMar 6, 2026UnknownInferred claimLow evidence strength

Achieved up to 90% efficiency gains in power plant work order planning, 40% efficiency improvements in finance processes like procure-to-pay, and improved customer experience and revenue transformation.

Normalized claim

Revenue: 40% increase

PwCMar 6, 2026UnknownInferred claimLow evidence strength

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

Customer: PwC
Provider: Microsoft
Maturity: Unknown
Linked source: PwC

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 3 of 3

1AI Agents
2Agent Orchestration
3Workflow Automation

PwC's agent OS accelerator, combined with Microsoft Azure AI Foundry, Copilot Studio, and Agent Service, provides an operating layer for AI agents that ensures governance, observability, and end-to-end workflow management in enterprise ecosystems.

Technologies

Azure AI Foundry Copilot Studio Azure AI Agent Service Model Context Protocol

Sources & evidence1

Evidence: Low35/100Evidence strength