Cisco and OpenAI Revolutionize Engineering with Codex, an AI Ally

Cisco and OpenAI: A Strategic Integration of Codex

Cisco has undertaken a significant transformation by integrating OpenAI's Codex into its software development processes. This extensive integration has allowed Cisco to make AI-native development an integral part of enterprise software creation. Today, over 95% of new AI features at Cisco are written by Codex, illustrating the major impact of this tool on technological innovation.

The use of Codex has enabled Cisco to increase its defect resolution capacity by 10 to 15 times through the Codex CLI tool, saving over 1,500 engineering hours per month. This increased efficiency is a major asset for a company that manages some of the most complex software systems in the world.

For decades, Cisco has been recognized for its ability to build and operate some of the most complex and critical software systems globally. With the emergence of generative AI, Cisco has leveraged its expertise to integrate these advanced technologies into real-world, demanding environments. This approach has already begun to transform how Cisco develops new products, notably with AI Defense, where Codex has significantly reduced the time required to develop critical features from several quarters to just a few weeks.

AI Defense: Enhanced Security Through Codex

Cisco's AI Defense project perfectly illustrates the impact of Codex in the security domain. AI Defense is a comprehensive AI security solution designed to protect against security and safety risks introduced by AI. Thanks to Codex, the Cisco team was able to write the majority of the AI Defense features, thus reducing the development time from several quarters to just a few weeks.

DJ Sampath, SVP/GM of AI Software and Platform at Cisco, emphasizes that features that would have taken several quarters to reach customers are now delivered in just a few weeks thanks to Codex. This work also reflects Cisco's broader role in advancing AI security. Cisco is one of the leading security organizations working with OpenAI's Daybreak initiative, which brings together OpenAI models, Codex, and security partners to accelerate cyber defense and continuously secure software. As part of this program, Cisco has regulated access to GPT-5.5-Cyber, a model for cyber defenders.

Cisco has also used Codex to help build their Defense Squad, an open-source tool that moved from ideation to the developer community in less than a week. This rapid development is a testament to Codex's effectiveness as an AI development tool.

Codex's Agency in Complex Codebases

Cisco already manages a mature engineering organization with several ongoing AI initiatives. What made Codex compelling was not just code completion or superficial automation, but its agency. Codex has demonstrated the ability to understand and reason through large interconnected repositories, work fluently in complex languages, and execute real workflows via autonomous compile-test-fix loops based on CLI. It also operates within existing review, security, and governance frameworks.

By collaborating directly with OpenAI, Cisco engineers have been able to provide feedback on the behavior of these capabilities in real-world environments, influencing areas such as workflow orchestration, security controls, and support for long-duration engineering tasks, all critical for enterprise use.

Optimizing Engineering Workflows

Once Codex was integrated into daily engineering work, teams began applying it to some of their most challenging and time-consuming workflows. For example, Codex analyzed build logs and dependency graphs across more than 15 interconnected repositories, identifying inefficiencies. This analysis led to approximately a 20% reduction in build times and over 1,500 engineering hours saved per month globally.

By using Codex-CLI, Cisco automated defect repair with iterative and agentic execution on large C/C++ codebases. What once took weeks of manual effort now concludes in just a few hours, providing a 10 to 15 times increase in defect resolution capacity and allowing engineers to focus on design and validation.

When the Splunk teams needed to migrate several user interfaces from React 18 to 19, Codex autonomously managed the majority of repetitive changes, compressing weeks of work into just a few days and enabling engineers to focus on judgment-based decisions. Ryan Brady, a senior engineer in Cisco's Splunk group, stated, “The biggest gains occurred when we stopped viewing Codex as a tool and started treating it as a team member. We use Codex to generate and track a plan document, allowing the review team to better understand both the process and the generated code.”

Shaping the Future of Codex for the Enterprise

Cisco has provided continuous feedback from real-world production use that has helped OpenAI accelerate Codex's readiness for large enterprises, particularly in areas such as compliance, long-duration task management, and integration with existing development pipelines.

For Cisco, this collaboration has established a repeatable model for adopting next-generation AI: deep technical partnership, real workloads, and alignment of leadership from day one. Today, Codex is used across several Cisco business units, enhancing productivity, code quality, and resolution time. Instead of measuring work solely by traditional effort metrics, teams are increasingly asking, “How long will this Codex execution take?” Brad Murphy, VP leading Cisco's Splunk engineering team, added, “Codex has become a significant part of our thinking about AI-assisted development and operations in the future.”