In the driver’s seat: How Google Conductor AI actually stays under control

The New Stack

by Adrian Bridgwater

February 25, 2026

AI-Generated Deep Dive Summary

Google has introduced an automated review feature for its Conductor AI extension, designed to give developers more control over their code. As part of Gemini CLI, Conductor helps create formal specifications alongside code, ensuring that developers can plan and review before writing a single line of code. The new update aims to generate post-implementation reports based on codebase guidelines, emphasizing the importance of maintaining human oversight in AI-driven development. However, while tools like Conductor and Anthropic's automated reviews are steps forward, they are not without limitations. Experts warn that relying solely on AI for code generation and review can leave gaps, such as vulnerabilities introduced by suggested packages or dependencies. Nigel Douglas from Cloudsmith compares this risk to a chainsaw without an off switch, stressing the need for human intervention at critical stages like pull requests. He advocates for keeping humans in the loop to verify AI-generated outputs, ensuring that trusted code is thoroughly reviewed before being integrated into projects. The industry's shift toward trusting AI-generated code is gaining momentum, with companies like Tabnine analyzing broader codebase contexts to provide actionable insights. Chris du Toit highlights that as AI takes on more engineering tasks, organizations must establish robust organizational intelligence layers to guide automated reviews safely at scale. This focus on reliability and context-aware evaluation marks a meaningful milestone in the evolution of AI development tools, aligning with growing concerns about safety and compliance in DevOps environments. For developers and teams, these advancements underscore the importance of balancing AI's efficiency with human oversight. While automated reviews streamline processes and enhance code quality, they cannot yet fully replace the critical thinking and domain expertise that humans bring to the table. As AI continues to evolve, the integration of human judgment into workflows will remain essential for ensuring that tools like Conductor AI stay under control and deliver trustworthy outcomes.

Verticals

devopscloud

Originally published on The New Stack on 2/25/2026