Codacy Launches AI Inventory to Combat Shadow AI in Engineering Teams
Key Takeaways
- ▸84% of developers use or plan to use AI coding tools, but organizations lack visibility into actual AI tool usage across repositories
- ▸91% of AI tools in organizations are unmanaged, creating ungoverned risk and security concerns that organizations cannot adequately assess
- ▸Shadow AI embeds itself in developer workflows and IDEs, making it invisible to traditional governance and security tooling, unlike shadow IT which leaves observable traces
Summary
Codacy has announced AI Inventory, a new tool designed to automatically detect and catalog AI tools, models, and integrations used across engineering repositories. The product addresses the "shadow AI" problem—the widespread, largely invisible adoption of AI coding tools by developers without organizational oversight or governance. According to Stack Overflow's 2025 Developer Survey, 84% of developers now use or plan to use AI coding tools, yet 91% of AI tools in organizations remain unmanaged, creating blind spots for security and compliance.
The shadow AI visibility gap has become increasingly critical as AI adoption accelerates at developer speed rather than procurement speed. Traditional approaches like surveys and manual audits have proven ineffective, yielding incomplete and quickly outdated information. AI Inventory uses code-level detection to identify which tools, APIs, and integrations are actually present in repositories, providing engineering teams with an accurate, real-time picture of their AI tool sprawl. This systematic visibility enables better governance, risk management, and informed decision-making around AI usage across development teams.
- AI Inventory provides code-level detection of AI models and tools, enabling organizations to maintain accurate, real-time catalogs of AI usage
Editorial Opinion
The launch of AI Inventory reflects a growing maturity in the AI governance space. As AI adoption outpaces organizational controls, tools that provide transparent visibility into actual usage patterns will become essential infrastructure for responsible enterprises. However, this product's success ultimately depends on whether organizations use the visibility it provides to implement thoughtful governance rather than reactive restriction—the goal should be enabling safe AI adoption, not preventing it.
