The rapid rise of AI-generated programming is transforming software development — but it is also creating a new challenge for engineering teams: reviewing the flood of machine-written code.
To address that problem, Anthropic on Monday introduced Code Review, an artificial intelligence system designed to automatically detect bugs and logical flaws in software before they enter production environments.
The tool, launched inside the company’s coding platform Claude Code, is aimed primarily at enterprise developers who are increasingly relying on AI assistants to generate large volumes of code.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
AI-powered coding assistants have rapidly changed the pace of software engineering. Developers can now describe a feature in plain language and receive working code almost instantly — a trend often referred to as “vibe coding.”
While the approach accelerates development, it also creates new risks. AI-generated code may contain subtle logic errors, security vulnerabilities, or poorly understood dependencies. At the same time, the volume of generated code has surged, increasing the number of pull requests that must be reviewed before deployment.
Pull requests — the standard mechanism developers use to submit code changes for review — have become a bottleneck for many engineering teams.
“We’ve seen a lot of growth in Claude Code, especially within the enterprise,” said Cat Wu, Anthropic’s head of product. “One of the questions we keep getting from enterprise leaders is: now that Claude Code is putting up a bunch of pull requests, how do I make sure those get reviewed efficiently?”
Code Review is designed to address that problem by automatically scanning submitted code and providing feedback directly within repositories hosted on GitHub.
Multi-Agent AI Architecture
The system operates using multiple AI agents running in parallel, each examining the codebase from a different analytical perspective.
One agent may focus on logical correctness, another on data flows, and another on historical patterns within the codebase. A final coordinating agent aggregates the findings, removes duplicate alerts, and ranks issues by severity.
The tool highlights problems using color-coded labels:
- Red for critical issues requiring immediate attention
- Yellow for potential problems that developers should review
- Purple for issues linked to legacy code or historical bugs
Unlike many automated code review systems that focus heavily on formatting or style rules, Anthropic designed Code Review to prioritize logical flaws.
“This is really important because a lot of developers have seen AI automated feedback before, and they get annoyed when it’s not immediately actionable,” Wu said. “We decided we’re going to focus purely on logic errors.”
The system also explains its reasoning step-by-step, outlining what the potential problem is, why it may cause issues, and how developers might fix it.
Enterprise Demand Drives New Tools
The launch reflects a broader shift in enterprise software development, where AI coding tools are rapidly becoming part of everyday workflows.
According to Anthropic, subscriptions for its enterprise products have quadrupled since the start of the year, and Claude Code’s annualized revenue run rate has surpassed $2.5 billion since its launch.
Large corporations — including Uber, Salesforce, and Accenture — are already using the platform, creating demand for tools that can manage the surge in AI-generated code.
Developer leads can activate Code Review across their engineering teams, allowing the system to automatically analyze every pull request submitted to a project.
The tool also performs basic security checks and can be customized to enforce internal coding standards or engineering policies. For deeper vulnerability analysis, Anthropic offers a separate product called Claude Code Security.
Running multiple AI agents simultaneously makes Code Review a computationally intensive service. Pricing is based on token usage — a common model in AI systems — and varies depending on the size and complexity of the code being analyzed.
Wu estimated that each automated review would cost between $15 and $25.
The company positions the product as a premium enterprise feature designed to handle the new scale of AI-driven development.
“As engineers develop with Claude Code, they’re seeing the friction to creating a new feature decrease,” Wu said. “But they’re also seeing a much higher demand for code review.”
The product launch arrives at an interesting moment for Anthropic. The company filed two lawsuits Monday against the U.S. Department of Defense after the agency classified Anthropic as a potential supply chain risk — a dispute that could affect its eligibility for certain government contracts.
As the legal battle unfolds, Anthropic appears to be doubling down on its rapidly expanding enterprise business, where demand for AI development tools continues to grow.
In that environment, automated code review may become a crucial component of the next phase of AI-assisted software development — helping companies manage the risks created by the very tools that are accelerating their productivity.