Powered by Azure Local • Azure Arc • Azure AI on Edge
Analysis
Ghidra ServerAzure AI VisionSemantic KernelAutoGen Agents
Execution
Hyper-V SandboxCheckpoint/ReplayNetwork CaptureUI Automation
Output
OpenAPI SpecsContainer ImagesTest SuitesGap Reports
Infrastructure
Azure LocalAzure ArcAKS HybridAzure AI on Edge
View Full Architecture Diagram →
// Analysis Layer

Intelligence that runs at the edge.

Ghidra Server

NSA's open-source reverse engineering framework running in headless mode. Performs automated binary analysis including function decompilation, call graph generation, string reference extraction, and import table analysis. Scales across multiple binaries concurrently using containerized workers.

Azure AI Vision

Screen recordings processed through Azure AI Vision on Edge for UI element detection, workflow annotation, and user interaction pattern recognition. Identifies buttons, forms, navigation patterns, and data entry workflows without requiring application source code.

Semantic Kernel

Microsoft's AI orchestration framework coordinates the analysis pipeline. Manages prompt chains for code understanding, specification generation, and business rule extraction. Runs against local LLM endpoints — no cloud API calls required.

AutoGen Agents

Multi-agent framework for collaborative code generation. Specialized agents handle different modernization tasks: API design, database migration, UI generation, test creation. Agents self-validate through peer review loops.

// Execution Layer

Sandboxed execution with full observability.

Hyper-V Sandbox VMs

Legacy applications installed and executed in isolated Hyper-V virtual machines. Clean-room approach: snapshot before installation, diff after. Each stage of analysis starts from a known checkpoint. VMs are disposable — no contamination between analysis runs.

Checkpoint & Replay

Full VM state captured at every phase boundary. Analysis can be rolled back, replayed, and branched. Deterministic execution enables reproducible results. Any failure point can be re-entered without starting from scratch.

Network Capture

All network traffic from sandbox VMs routed through a transparent proxy. HTTP/HTTPS, SQL, SMTP, and custom protocols captured and decoded. Identifies every external dependency the application requires — including undocumented ones.

UI Automation

Programmatic interaction with legacy application UIs. Click-path automation exercises discovered workflows and validates behavior. Captures input/output pairs for regression test generation. Handles Win32, WinForms, WPF, and web-based interfaces.

// Output Layer

Artifacts you own and maintain.

📋

OpenAPI Specs

RESTful API definitions generated from discovered behavior. Versioned, documented, ready for consumer integration.

📦

Container Images

Docker/OCI images with Helm charts and AKS manifests. Multi-stage builds, security-scanned, production-ready.

🧪

Test Suites

End-to-end, integration, and regression tests derived from captured workflows. Continuous validation built-in.

📊

Gap Reports

Comprehensive documentation of what was automated, what requires manual intervention, and what was excluded.

// Infrastructure Layer

Azure Local: the foundation for sovereign compute.

Why Azure Local?

The modernization pipeline requires significant compute — GPU acceleration for AI workloads, large memory for binary analysis, and fast storage for VM checkpoint/replay. Azure Local provides enterprise-grade hyperconverged infrastructure managed through Azure Arc.

Organizations get cloud-native management and tooling (Azure Portal, ARM templates, Azure Policy) while keeping all data and processing physically on-premises. No data egress. No cloud dependency for core pipeline execution.

  • Azure Local — Hyperconverged infrastructure with Azure management plane. GPU passthrough for AI workloads (Nvidia A2/T4).
  • Azure Arc — Unified control plane across on-premises and cloud resources. Policy, monitoring, and update management from Azure.
  • AKS Hybrid — Kubernetes on Azure Local for container orchestration. Runs the modernized output alongside the analysis pipeline.
  • Azure AI on Edge — Local AI model inference. Vision, speech (Whisper), and language models running on GPU nodes without cloud connectivity.

Your legacy. Your code. Your infrastructure.

Request Assessment