AI can now understand and transform big enterprise codebases

MCP server for CAST Imaging unlocks architectural context for LLMs

PARIS and NEW YORK, Nov. 20, 2025 (GLOBE NEWSWIRE) -- CAST, a leader in software mapping and intelligence technology, today announced a major advance in how AI can understand, improve, and transform large, complex enterprise applications. With the general release of its MCP server for CAST Imaging, AI agents can now access the precise architectural context information that they require to work on existing code. CAST Imaging generates detailed maps of an application’s internal structures, and the MCP server allows AI agents to access those insights. The server, introduced in beta in August, is now available to all CAST Imaging users.

“To make the right changes to code, AI needs a complete and correct map of the software it’s working on,” said Olivier Bonsignour, Head of R&D at CAST. “The MCP server delivers it by connecting the AI to CAST Imaging, which generates the information. It’s the difference between AI assuming and knowing.”

Impact on companies
Through this new architectural understanding, AI can now perform missions previously out of reach. These include remediating technical debt, modernizing applications in-place or on-cloud, replacing frameworks or databases, and analyzing the impact of changes. Workstreams that once required months can now be completed in minutes, materially reducing the cost of major initiatives and making previously impractical efforts achievable.

How it works
AI models struggle to fully understand enterprise-scale applications. As probabilistic systems, Large Language Models (LLMs) cannot accurately deduce the relationships between objects, data, and frameworks that define existing custom enterprise software. Their context windows are further challenged by applications containing more than a million lines of code and those built from multiple types of programming languages and technologies. CAST Imaging closes this gap by feeding the AI the deterministic information it needs to comprehend and work on existing code.

Loaded with insights from CAST Imaging, the MCP server shares the application’s internal architecture, including code objects, data structures, dependencies, frameworks, and cross-application relationships, with the AI. Delivered as a Docker container or a Windows extension, it can run on-premises or in the cloud, allowing AI agents to ‘see’ and reason about software at any level of detail.

CAST Imaging understands applications comprised of any mix of over 150 technologies. It semantically analyzes applications, mapping every explicit and hidden link between code and data into architectural graphs. The MCP server then makes these maps accessible to external AI agents through a secure interface, providing information such as:

1. Application statistics
2. Cross-app dependencies
3. Internal architectural graphs
4. Transactions, data graphs, packages
5. Object details, interactions, call graphs
   

MCP server access is available to all CAST Imaging users. Activation instructions and documentation can be found at https://www.castsoftware.com/mcp.
  
About CAST
Businesses move faster using CAST to understand, improve, and transform their software. Through semantic analysis of source code, CAST generates dashboards and 3D maps for executives, technologists, and AI to navigate inside individual applications and across entire portfolios. This intelligence enables companies to steer, speed, and report on initiatives such as technical debt, modernization, and cloud. As the pioneer of the software intelligence field, CAST is trusted by the world’s leading companies and governments, their consultancies and cloud providers. See it all at castsoftware.com.

For more information, please contact David Rosen at d.rosen@castsoftware.com.

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.