The rise of agentic AI from strategy to security in the workplace

A growing chorus of industry leaders from KPMG and IBM to Stanford and Google are aligning around one message in 2025: Agentic AI is no longer experimental, it’s operational. A sweeping wave of new publications reflects how fast enterprises are adapting to this shift and what’s required to do it safely, securely, and strategically.

A Strategic Pulse Across the Enterprise

• KPMG’s Q2 2025 AI Pulse Survey confirms that over 70% of enterprises are now piloting or deploying agent-based AI systems to augment workflows, decision-making, and customer operations.

• BCG’s “AI at Work” underscores that the frontier is no longer AI-assisted productivity tools, but agent ecosystems that continuously learn and act within enterprise environments.

• KPMG: AI Quarterly Pulse Survey: Q2 2025 (Doc)

• BCG: AI at Work (Doc)

Security and Governance Take Center Stage

• Google’s “Introduction to AI Agent Security” and Stanford’s “Future of Work with AI Agents” both emphasize a critical turning point: without robust governance, role separation, and context control, agentic systems can amplify misinformation, bias, and operational risk.

• Thomson Reuters’ Agentic AI 101 frames this shift as a legal and compliance issue, not just a technical one.

• Google: An Introduction to AI Agent Security (Doc)

• Google: Guide for using AI at work (Doc)

• Stanford University: Future of Work with AI Agents (Doc)

• Thomson Reuters: Agentic AI 101 (Doc)

Building Responsibly: Toolkits and Benchmarks

• OpenAI’s “Practical Guide to Building Agents” provides hands-on frameworks for orchestrating task agents and tool usage securely within enterprise constraints.

• ServiceNow’s Enterprise AI Maturity Index 2025 segments organizations by readiness, indicating most firms remain stuck at “augmented workflows,” far from truly autonomous decision-making.

• OpenAI: A Practical Guide to building Agents (Doc)

• ServiceNow: Enterprise AI Maturity Index 2025 (Doc)

Sector Spotlight: Financial Services

• IBM’s “Agentic AI in Financial Services” explores use cases such as fraud detection and intelligent advisory systems while stressing the need for real-time monitoring, explainability, and token-level auditing.

• IBM: Agentic AI in Financial Services (Doc)

A New Operating Model Emerges

• Capgemini’s “Rise of Agentic AI” posits that agent ecosystems represent a new digital operating model, not a technology stack, but a cognitive infrastructure. The challenge: aligning this agility with legacy controls, human oversight, and business outcomes.

• 1. Capgemini: Rise of Agentic AI (Doc)