Enterprise AI agents move from experiments to operating model
Enterprise AI agents are software entities that use large language models, enterprise data and business rules to perform tasks autonomously or alongside people across workflows, and they are now shifting from limited pilots to production-scale deployment through trusted consulting and IT service provider networks that bring governance, security and organizational change capabilities into a single operating model. KPMG’s expanded relationship with Microsoft shows how this shift is unfolding. The firm is standardizing on Microsoft Agent 365 to control how AI agents are deployed, monitored and updated, while rolling out Microsoft 365 Copilot more widely as a secure, enterprise AI platform for its 276,000 professionals. Agent 365 strengthens KPMG’s Workbench ecosystem with a centralized control plane for AI agents spanning systems, data and processes. Together with the KPMG Trusted AI framework, these tools move enterprise AI agents from isolated experiments to a repeatable, governed model clients can scale.
KPMG and Microsoft build a trusted enterprise AI agent framework
KPMG is using Microsoft 365 Copilot and Agent 365 to create a structured, AI governance framework for its own teams and its global client base. With Copilot embedded into everyday tools, professionals in audit, tax and advisory can automate document creation, analysis and collaboration while keeping firm data and intellectual property protected inside the Microsoft Copilot enterprise environment. Agent 365 adds centralized visibility over AI agent deployment, giving KPMG the ability to set policies, define ownership and manage lifecycle control for agents running across workflows. According to Lisa Heneghan, Global Chief Digital Officer at KPMG, scaling AI “requires strong foundations in governance, visibility and accountability” so AI becomes a trusted part of culture rather than a set of disconnected tools. By packaging these practices into its Trusted AI framework, KPMG is turning its own deployment playbook into a client-ready model for enterprise AI agents.
Atos bets on secure agentic AI for every employee
Atos Group is taking an all-in approach to secure agentic AI, deploying Microsoft 365 Copilot to all 56,000 employees across 54 countries as part of Microsoft 365 E7. This foundation brings together Microsoft 365 Copilot, the E5 security and compliance stack and Microsoft Agent 365 into a unified platform, reinforced by Entra Suite plus Defender, Intune and Purview. Atos is already managing a population of 19,000 AI agents through Agent 365, using the same administration and security workflows its IT teams run today. It is also using Microsoft Copilot Studio and Microsoft Foundry to design, build and operate agents across internal IT, business functions and client projects, feeding into its Sovereign Agentic AI studios. Frédéric Aubrière, Group Chief Digital & Information Officer, calls this “the most significant technology investment in our people that Atos has made in a generation,” signaling that AI assistants are becoming standard equipment, not optional tools.
Security, governance and trust as the pillars of AI agent deployment
Both KPMG and Atos are framing AI agent deployment around security, governance and trust rather than individual features. Microsoft Agent 365 plays a central role as a control plane for observing, governing and securing AI agents, whether they act on behalf of users or operate with their own credentials. In practice, that means centralized policies, consistent risk and compliance checks and clear ownership for agents across their lifecycle. At KPMG, the Trusted AI framework adds ethical and regulatory guardrails so clients can introduce agents into complex domains like audit or regulated industries with documented oversight. Atos complements this by basing its rollout on Microsoft 365 E7, Entra, Defender, Intune and Purview, so AI agents inherit existing identity, device and data protections. This alignment lets organizations scale secure agentic AI without building parallel security stacks, reducing friction and keeping AI initiatives auditable and accountable.
Partnership-led models accelerate enterprise AI agent adoption
The common thread in both announcements is the use of established consulting and IT service provider relationships to speed AI agent deployment. KPMG and Atos already sit inside clients’ core transformation programs, giving them insight into processes, data and risk requirements that AI agents must respect. By standardizing on platforms like Microsoft 365 Copilot enterprise and Agent 365, these partners can offer repeatable blueprints instead of one-off pilots, shortening implementation timelines and improving consistency. Clients benefit from reference architectures, pre-defined AI governance frameworks and shared lessons from large internal deployments. Atos is also reinforcing joint go-to-market initiatives with Microsoft so customers in regulated industries can see how secure agentic AI works in a live, enterprise context. As more service providers follow similar patterns, enterprise AI agents are likely to spread not through isolated innovation labs, but through scaled, partnership-led operating models.
