PwC’s Agentic Enterprise Strategy: Scaling AI for Real Outcomes, Human Expertise, and Competitive Advantage

PwC’s AI Imperative: A Strategic Moment for Enterprise Change

Artificial intelligence is no longer a future possibility; it is actively transforming enterprise operations today. With economic uncertainty and the rise of generative AI, companies face a clear message: staying competitive depends on the ability to apply, govern, and scale AI effectively.

At a recent event in San Francisco, leaders from PwC, including U.S. Chief AI Officer Dan Priest and Global and U.S. Commercial Technology and Innovation Officer Matt Wood, outlined how the firm is responding. Their approach focuses on fundamentally reinventing how business performance, knowledge work, and human expertise combine with intelligent automation. Priest emphasized that AI is an amplifier of expertise, not a replacement. This principle guides PwC’s AI strategy, which prioritizes delivering operational value at scale, responsibly transforming legacy systems, and embedding AI to augment people rather than substitute them.

What Enterprises Demand

Enterprises today aren’t seeking more AI tools; they want clear outcomes. Their goals include modernizing legacy applications without unacceptable risks, breaking down siloed systems, and driving continuous improvements from existing infrastructure. They also want to deploy secure AI agents that perform real work across business processes.

More importantly, enterprises aim to embed AI in ways that balance innovation with governance, oversight, and trust. According to PwC’s May 2025 AI Agent Survey, 88% of senior executives plan to increase AI budgets within the next year, and 79% already use AI agents. Yet, only 35% have broadly adopted agents, and fewer than half have redesigned work fundamentally to incorporate AI.

PwC’s Strategic Response: Architecting AI for Real Outcomes

To meet client needs, PwC created a multi-layered approach centered on agent OS, an enterprise-grade operating system for intelligent agents. This orchestration framework operates across all major cloud platforms—Amazon Web Services, Microsoft Azure, Google Cloud—and integrates with model vendors like OpenAI and Anthropic, as well as enterprise applications including Salesforce, Oracle, SAP, and Workday.

Agent OS acts as a secure switchboard, connecting, coordinating, and governing AI agents at scale. A key feature is its integration with the Model Context Protocol (MCP), an open-source standard from Anthropic. MCP enables agents to securely access external tools and data, which is vital as organizations move from pilot projects to full production deployments. It supports consistent and reusable agent-to-tool interactions, accelerating development and simplifying deployment.

Security is baked in from the start, with static code analysis, credential vaulting, and identity-based authorization. This addresses a top concern among executives—cybersecurity, cited by 34% of those surveyed as a primary barrier to realizing AI agent value.

From Concept to Reality: Enterprise-Wide AI Transformation

PwC’s AI strategy is active and measurable. Internally, the firm has deployed over 250 AI agents and more than 12,000 custom GPTs, logging upwards of 31 million generative AI interactions. More than 95% of U.S. employees have participated in AI skill-building, contributing over 460,000 training hours and completing 50,000+ hands-on sessions.

These efforts have yielded significant productivity gains:

• Software development productivity improved by 20% to 50%
• Finance productivity rose between 20% to 40%
• Marketing productivity increased from 20% to 30%

Clients have seen similar results:

• Southwest Airlines modernized its crew management platform with generative AI, accepting 90% of AI-generated user stories and halving planning time.
• Wyndham Hotels & Resorts sped up brand standard reviews by 90% and cut contact center handling time by 50%, enhancing guest and owner experiences.
• Cross Insurance automated document classification and quote generation, reducing quote handling costs by 20% and fully managing 50% of inquiries via AI agents.

Survey data reinforces these outcomes: Among AI agent users, 66% report increased productivity, 57% see cost savings, and 54% note improved customer experience.

Agent-Powered Performance: Real-Time Sense, Think, and Act

One of PwC’s innovative offerings is Agent Powered Performance, integrated with agent OS. This embeds AI agents directly into core enterprise systems like ERP and CRM. Agents follow a “Sense, Think, Act” model: they monitor value leakage, evaluate predictive models and benchmarks, and execute improvements inside operational tools.

For instance, a finance agent can detect and correct billing anomalies before they impact ledgers, while a supply chain agent may reroute shipments to avoid cost spikes. This approach ensures continuous value delivery, with early adopters reporting measurable improvements within the first quarter and compounding benefits over time.

Addressing the Human Factor

Technology maturity is not the main challenge; adoption and organizational readiness are. The AI Agent Survey highlights that while cybersecurity and costs are concerns, deeper barriers include employee adoption (14%) and readiness for organizational change (17%). Only 42% of companies have redesigned workflows around AI agents, despite 50% expecting their operating models to change dramatically within two years.

PwC tackles this gap with a comprehensive workforce strategy. With over 3,200 AI champions and extensive training programs, the firm keeps human expertise central to AI transformation. Agents are built to augment human workers, freeing staff from low-value tasks and enabling focus on strategic, high-impact work. As Matt Wood pointed out, “AI is driving the cost of programming down to zero, but the value of engineering and business problem-solving is going up.”

Looking Ahead: Multi-Agent Orchestration and Competitive Edge

PwC views multi-agent systems as the next key architecture for enterprise AI. These systems allow various agents—planners, executors, auditors—to collaborate toward shared business goals. Through agent OS, PwC plans to enable these complex designs in real-world client environments.

The market is still emerging: fewer than half of surveyed companies use agents to redesign operating models or develop new products. However, executives recognize the potential, with 73% believing AI agents will deliver a competitive advantage within the next year.

Conclusion

PwC is not just advising clients on AI adoption but helping operationalize it with measurable results and cultural alignment. Their technologies, like agent OS and secure MCP integration, combined with outcome-focused solutions such as Agent Powered Performance, provide a practical blueprint to scale AI confidently and quickly.

For organizations committed to transforming work—beyond simple digitization—PwC offers a unique blend of engineering discipline, domain knowledge, partnerships, and enterprise platforms. This positions their clients to build truly agent-driven enterprises and secure sustainable advantages in a changing business landscape.