Atos launches Genesis Intelligence and Codex for controlled enterprise AI

Atos announced Genesis Intelligence on October 26, bringing data management, AI capabilities and operational controls into an enterprise suite. It also introduced Atos Codex, a platform intended to connect AI-generated insights with actions in critical operations. The company is targeting organizations that handle sensitive information or run services where interruptions matter, including transport operators, industrial businesses and public institutions.

Atos Codex stands for Center of Operations and Data EXcellence. Its proposed foundation is a shared representation of an organization’s activities: the relevant people and systems, their relationships, applicable rules, permissions and available actions. That gives users and AI a common business context. Atos says the organization determines how decisions become actions and how much human control remains, rather than treating a recommendation as permission to execute it.

The suite includes data preparation and traceability, model selection and cost management, a factory for smaller locally deployable language models, and an enterprise portal for customizable agents. Atos describes deployment across public or sovereign clouds, customer premises and disconnected environments. The launch announcement does not provide a general price list, deployment timetable or independently measured customer outcomes.

The services behind that proposition were taking shape earlier in the year. On March 12, Atos announced Sovereign Agentic Studios, a delivery model spanning four countries and supported by ten global delivery centers. Its account identified work in Birmingham and Texas, alongside capabilities in France and Germany. It positioned security, oversight and the ability to choose where data and models operate as part of delivering agents into everyday business operations.

A May 20 company analysis explains the problem those controls address. Atos identified risks including data exposure, manipulated model outputs and unauthorized access, while arguing that opaque decisions make errors harder to find. Its proposed response combines assessments, policies, testing, identity controls and monitoring. This is the supplier’s account of its security approach, not evidence that packaging those elements together eliminates every failure mode of an AI system.

That analysis also described earlier consulting assignments. An unnamed public-sector organization testing Microsoft 365 Copilot received a layered assessment of software, AI-specific risks and privacy issues. Another anonymous technology customer received a review of a generative AI application platform and recommendations to address weaknesses. These accounts show the type of implementation work Atos is bringing into the new offering, although the blog does not identify them as deployments of Genesis Intelligence.

The March Studios announcement adds a partnership dimension. Atos said it was working with Poolside on enterprise agents that could run on customer-owned infrastructure, combining the partner’s platform with its own operating controls. Its Scaler accelerator also connected customers with companies covering process assessment, workforce support and agent security. The October suite builds on an existing services and partner network.

The earlier announcement named the UK’s Department for Environment, Food and Rural Affairs as a Studios customer. The department described modernization work and further AI use cases undergoing structured assessment. That relationship predates the Genesis Intelligence launch and concerns the Studios engagement. It provides context for Atos’s focus on critical public services while keeping the newly announced suite separate from customer results reported months earlier.