The challenge
A large federal agency depended on a custom enterprise financial management system for budget formulation and execution. The platform was mission-critical, used across the organization, and deeply embedded in core workflows.
When the external entity that originally built the application abruptly transferred ownership, the agency inherited a system no one fully understood. Roughly 20% of the source code was redacted. Classification names were obscured. Documentation was sparse to nonexistent. Inline comments were virtually absent. Operating the application—and planning for its future—meant navigating a black box system with significant technical risk.
The agency faced a pressing question: How do you safely modernize what you can't see?
The solution: AI augmented modernization with human-led oversight
CGI applied its AI-powered IT modernization offering—an approach purpose-built to illuminate complex, undocumented legacy systems.
Rather than relying on months of system analysis and generic AI code generation, the model mirrors a full software engineering organization, including:
- Program management and structured delivery governance
- Security and risk oversight
- Sprint planning and iterative mission definition
- Continuous SME feedback loops
- Specialized agentic AI teams performing tightly scoped analysis tasks
Each AI agent conducted targeted investigations into the legacy codebase. Findings were visualized in a dynamically evolving knowledge graph, accessible through an interactive UI. With every iteration, patterns sharpened, architectural relationships became clearer, and system behavior was mapped with increasingly deterministic accuracy—thus minimizing and mitigating vulnerability risk up front.
Subject matter experts from the agency and CGI collaborated throughout, validating discoveries and accelerating shared understanding.
The approach: from opaque code to a living knowledge framework
- The AI-assisted environment rapidly transformed the legacy application from an opaque asset into a navigable, extensible system. Key capabilities included:
- Automated reverse engineering of code structures and data flows
- Visualization of dependencies and architectural components
- Iterative enrichment of a deterministic knowledge layer
- Role-specific insights tailored for developers, analysts, and operations teams
- Early risk and vulnerability identification
The result was not simply documentation—it was a modernization-ready blueprint that the agency could use immediately.
The outcomes: months of engineering insight delivered in hours
What would ordinarily require months of intensive effort from a team of full-time engineers was completed within hours. The team delivered a comprehensive documentation hub, along with role-based quick start guides designed specifically for analysts, developers, architects, and operations staff. They also produced fully reverse-engineered API specifications, security attack landscape, and a consolidated, searchable knowledge base that provides end-to-end architectural visibility across components and data flows. With this clear and validated understanding of its system, the agency now benefits from a significantly reduced technical risk profile, a solid foundation for either phased or full modernization, and the confidence needed to plan future enhancements and operational improvements.
Why it matters: a new path to modernizing legacy systems
Historically, federal agencies burdened by undocumented or partially inaccessible legacy applications had limited options—accept the risk, rebuild from scratch, or undertake long, costly discovery projects.
CGI’s AI-powered IT modernization approach solves this challenge.
By applying a controlled, agentic AI analysis layer with strong human governance, agencies can illuminate systems previously considered too complex, too fragile, or too opaque to modernize.
The outcome is a realistic, accelerated, and low-risk path from technical debt to transformation—unlocking mission value that was previously out of reach.
