Cisco Targets 70% AI-Written Products
Jeetu Patel, Cisco's President & CPO, revealed the company's AI Defense product features 100% AI-written code and set a goal for 70% of Cisco products to be AI-written by 2027. Patel outlined a new mental model for product teams focusing on judgment, instinct, outcomes obsession, unit economics, safety, and managing AI agents. The post generated significant engagement with 37 likes, 9 reposts, and 4,748 views.
- To achieve its AI-driven development goals, Cisco is transitioning to a "spec-driven" model where human engineers focus on writing detailed specifications and requirements, which are then fed to AI agents to generate the code. This approach is a significant workflow change from traditional agile development. - The shift to AI-written products involves a new team structure, moving from a team of eight human engineers to a model of three humans working with five AI agents. Jeetu Patel claims this new structure can triple the team's output. - By the end of 2026, Cisco aims to have at least half a dozen products where 100% of the code is written by AI, with human developers primarily serving as reviewers and spec writers. - This strategy is part of a broader "AI-first" cultural shift at Cisco, which emphasizes that every employee, from engineering to marketing, should integrate AI into their core workflows as a "digital co-worker." - To address the security risks of AI-generated code, which can include vulnerabilities and insecure defaults, Cisco has open-sourced its internal framework, Project CodeGuard, to integrate security rules into AI coding workflows. - In the broader industry, other tech giants are also heavily integrating AI into their coding processes. As of early 2025, Google reported that AI systems were generating over 25% of the new code for its products. - Analysts view Cisco's aggressive AI strategy as a key factor for future growth, with firms like Bank of America and Morgan Stanley raising their price targets for the company's stock, citing strong AI-related orders. However, some analysts note potential short-term pressure on profit margins due to the high cost of memory and other hardware required for AI infrastructure. - The AI Defense product, the first to be fully coded by AI, is designed to secure the entire AI lifecycle, from discovering a company's AI assets to detecting vulnerabilities and protecting against runtime threats like prompt injections and data leakage.
