Atos expands Microsoft partnership to deploy secure agentic AI across 56,000 employees and client work

Atos Group and Microsoft announced an expanded strategic collaboration aimed at scaling what Microsoft calls “secure agentic AI” across Atos’s workforce and its client delivery. In a joint statement, the companies said Atos will deploy Microsoft 365 Copilot broadly inside the firm and use Microsoft’s tools to build and operate AI agents for internal IT and business teams, while also bringing the same “playbook” to client engagements.

Atos said the rollout will reach all 56,000 employees across 54 countries. The deployment is described as being built on Microsoft 365 Copilot E7, which the companies characterize as a single unified solution that brings together Microsoft 365 E5, Microsoft 365 Copilot and Agent 365, powered by “Work IQ.” Atos also said the rollout is augmented by Microsoft Entra Suite and security capabilities including Microsoft Defender, Intune and Purview.

“Agentic” in this context refers to AI software that can take actions toward a goal, rather than only generating text. Microsoft’s announcement frames the expansion as an effort to operationalize that kind of AI for day-to-day work while keeping it governed and protected by enterprise identity, endpoint management and data protection tooling.

Atos also positioned the program as a validation milestone for Microsoft’s platform in France and Europe. The statement says Atos will become the first French global system integrator to deploy Microsoft 365 Copilot, and one of the first France-headquartered companies to operate fully on Microsoft 365 E7 from general availability.

Beyond employee productivity, Atos said it will use Copilot Studio and Microsoft Foundry to design, build and operate AI agents for Atos IT, Atos business functions and for Atos clients. Microsoft generally describes Copilot Studio as a toolset for creating and managing copilots and connected experiences, while Foundry is aimed at building and running data and AI solutions. The companies said Atos will bring these efforts together under a single “coherent operating model.”

The expansion comes at a time when large services firms are racing to move beyond pilot projects into sustained deployments of AI that can be embedded in processes and governed at scale. For IT providers, the added attraction is that a standardized internal platform can also be packaged into repeatable client offerings, reducing friction in delivering AI transformations across industries and regulatory environments.

What is not disclosed in the announcement is any financial impact, timeline beyond the stated global employee rollout, or specific governance targets such as measured reductions in cycle time, error rates, or incident rates. The companies also did not provide details on which workstreams or job functions will receive the earliest agent capabilities, or how Atos plans to segregate client data when agents are used in customer environments.

In the months ahead, the key developments to watch are how Atos defines “secure” in practice for agentic workflows, whether it publishes implementation details for Copilot E7, and how quickly its Copilot Studio and Foundry-based agent builds translate into customer deployments. The announcements suggest a shift from experimentation to operating AI agents as part of normal delivery, but the real test will be performance, security outcomes and repeatability across diverse client settings.