Dell brings OpenAI Codex into governed enterprise development environments

Dell Technologies has introduced a reference architecture with OpenAI that brings Codex to Dell AI Factory, combining OpenAI models with customer-managed development environments and governed enterprise data. The design is intended to let Codex investigate issues, propose code changes and test fixes while existing access controls remain in place.
- Dell introduced a Codex reference architecture with OpenAI at SC26.
- Red Hat OpenShift Dev Spaces provides isolated, centrally managed workspaces on Dell infrastructure.
- Dell AI Data Platform supplies governed security, operational, business and audit context.
- Codex can investigate vulnerabilities, propose and test fixes, and support developer verification.
- The architecture also targets incident response, modernization, analytics and workflow automation.
Codex works inside a customer-controlled environment
The architecture places Codex in centrally managed workspaces provided by Red Hat OpenShift Dev Spaces and running on Dell infrastructure. These workspaces contain source code, dependencies and approved development tools, while repository operations, terminal commands, code changes and testing take place locally in the customer-controlled environment.
Dell says the design keeps code, tool execution and enterprise data on Dell infrastructure. Workspace policies determine which repositories, applications and services an agent can access. Developers remain involved by reviewing proposed changes, approving modifications and controlling how they move through the software delivery process.
Enterprise context comes through Dell AI Data Platform
Dell AI Data Platform supplies operational, business and analytical context for Codex. Dell plugins connect Codex to the platform's engines over MCP. The Data Search Engine provides security events and alerts, the Data Analytics Engine provides aggregated operational, business and audit context, and Data Orchestration can automate how relevant data is ingested and prepared.
The architecture adds several layers of control around the agent. It runs on Red Hat OpenShift and OpenShift Dev Spaces on Dell PowerEdge servers, PowerScale storage and PowerSwitch networking. Network policy denies egress except to approved destinations, each developer's namespace is isolated, the workspace pod runs as non-root and the built-in Codex sandbox restricts the commands the agent can run. Dell AI Data Platform plugins apply each engine's own access policies when Codex reaches enterprise data.
A security alert becomes a developer-reviewed fix
Dell's demonstration begins when a security rule in the Data Search Engine detects suspicious application behavior and raises an alert through Kibana. Codex does not create the alert. Instead, it uses evidence from the Dell AI Data Platform, combines the security event with business and audit context, and guides an investigation of the related code.
The workflow has five stages: detect, investigate, trace, remediate and verify. Codex examines the importer service, identifies a vulnerable code path, proposes and tests a focused change, and presents the result for developer review. When the original attack is repeated, the application blocks the request, the Search Engine records the attempt and the Analytics Engine confirms the blocked outcome in audit data.
The model extends beyond security remediation
Dell presents the security example as one use of a broader operating model for enterprise AI agents. According to Dell, the same architecture can support incident response, application modernization, operational investigation, business analytics and compliance reporting, as well as code review, quality assurance, data analysis and workflow automation.
The announcement matters because it addresses where an agent runs, where code is held, how it obtains enterprise context and whether existing governance applies. Dell's design is aimed at organizations that want Codex to act across approved repositories, operational data, documentation and engineering systems without moving their development environment or operational data into a separate coding platform outside of Codex.
Dell Technologies: Dell Brings Codex to Dell AI Factory with OpenAI

