For decades, the Department of the Air Force has run mission-critical workloads on infrastructure and code that has long outlived its original architectural assumptions. COBOL, Ada, mainframe-bound business logic, monolithic ground-control systems, and bespoke C2 stacks built across multiple acquisition cycles now collectively represent both an operational lifeline and a strategic liability. Every dollar spent sustaining this technical debt is a dollar not spent on readiness, lethality, or speed-to-decision. The conventional answer of a multi-year lift-and-shift modernization programs has consistently underdelivered. AI changes the math.

This session presents a field-tested playbook for using AI as the catalyst, not the replacement, for legacy modernization, drawn from Insight's work modernizing complex application portfolios across federal civilian, defense, and Fortune 500 environments. We will walk through a four-phase model that pairs AI-assisted code intelligence with disciplined cloud-native engineering: 

• AI-driven discovery and dependency mapping of legacy estates where source documentation is incomplete or lost
• Automated code comprehension and refactoring using LLM-based tooling to translate, document, and decompose monoliths into bounded services
• AI-augmented migration to FedRAMP- and IL-aligned cloud-native environments with embedded DevSecOps
• Continuous modernization loops where AI surveils technical debt and recommends remediation as a steady-state practice rather than a program of record.

Attendees will see concrete examples of how this approach has compressed timelines and reduced risk in environments analogous to those facing the Business and Enterprise Systems (BES) Directorate and Space Force ground enterprise. Reference patterns include AI-assisted COBOL-to-.NET translation in regulated financial services, automated decomposition of decades-old supply-chain systems for a major retailer, and the application of LLM agents to generate test scaffolding and security baselines for code that was previously considered "untouchable." We will translate those lessons into Air Force–relevant contexts: refactoring satellite ground control software where determinism and safety cannot be compromised; modernizing C2 platforms where mission assurance is non-negotiable; and reducing the sustainment tail on BES financial, logistics, and personnel systems.

The session will also confront the honest constraints. AI-assisted modernization is not magic, it requires human-in-the-loop validation, mature reference architectures, robust evaluation harnesses, and acquisition vehicles that can absorb iterative delivery. We will share Insight's framework for measuring modernization ROI in mission terms (readiness uplift, time-to-deploy, sustainment cost per capability) rather than in lines of code translated, and the governance patterns that keep AI-generated artifacts auditable and trustworthy.

Key takeaways:

• A reusable, four-phase AI-assisted modernization model adaptable to DAF environments
• Practical guidance on tool selection, evaluation, and human-in-the-loop controls for AI-generated code
• Mission-aligned metrics that connect modernization investment to warfighter readiness
• A candid view of what AI can and cannot accelerate in classified and safety-critical estates
• An adoption roadmap that meets BES, Space Force, and MAJCOM teams where they are today