AI company Anthropic has introduced a new tool designed to review large volumes of AI-generated code automatically. The feature, called Code Review, is launching within the company’s AI coding platform Claude Code.
The tool aims to help development teams detect bugs, logic errors, and potential security issues before code is merged into production systems.
AI Coding Creates New Development Challenges
AI-powered development tools have made it possible for programmers to generate large amounts of code using simple instructions written in natural language. While this approach speeds up development, it also increases the risk of hidden bugs and poorly understood code.
Anthropic’s new Code Review system acts as an automated reviewer designed to identify these issues early in the development process.
Built for Enterprise Teams
The product is targeted at large enterprise organizations already using Claude Code, including companies such as Uber, Salesforce, and Accenture.
Many of these companies are seeing a surge in pull requests as AI tools generate more code than ever before.
Integration With GitHub
Once activated, the Code Review feature integrates with GitHub and automatically reviews pull requests. The AI analyzes the code and leaves comments explaining potential problems and recommended fixes.
Issues are categorized using a color-coded severity system, helping developers prioritize which problems need attention first.
Multi-Agent AI Architecture
Anthropic says the system relies on multiple AI agents running simultaneously. Each agent analyzes the codebase from a different perspective before a final agent consolidates the results and prioritizes key findings.
Pricing and Availability
The Code Review tool uses token-based pricing, with Anthropic estimating that each automated review will cost between $15 and $25 depending on code complexity.
The feature is currently available in research preview for Claude for Teams and Claude for Enterprise customers.
Read more:
Latest AI News | Startup Stories | Technology Updates
