Best Practice

Agentic AI for Legacy Modernisation

Core Reply’s governed modernisation framework helps organisations recover legacy knowledge, shape the right target architecture, and modernise step by step without compromising continuity.

Modernisation Starts with Knowledge Recovery

Organisations managing mission-critical systems face the complex challenge of innovating rapidly while maintaining operational stability, compliance, and service continuity. AI can significantly accelerate analysis, but modernisation only succeeds when it is anchored to a rigorous understanding of the existing system. For Core Reply experts, legacy modernisation is not a code-rewriting exercise: it starts with reconstructing operational knowledge, then translating that knowledge into architectural choices, phased roadmaps, and controlled execution.

AI Can Accelerate Understanding of Legacy Systems

AI tools can materially accelerate reverse engineering by extracting and analysing legacy artefacts such as COBOL source code, Job Control Language (JCL), scheduler configurations, database scripts, interfaces, and copybooks. They help identify dead code, reconstruct dependencies, map data lineage, and generate first-pass technical documentation, reducing the manual effort required to understand complex landscapes.

However, code conversion is only one piece of the puzzle. A successful modernisation programme must also address integration with external systems, batch architecture and cut-off windows, achieving performance levels comparable to the mainframe, distributed transaction design, and how to preserve unit-of-work semantics in a distributed environment.

The Strategic Role of Agentic AI

In Core Reply’s approach, Agentic AI operates within a governed framework that turns fragmented legacy artefacts into a versioned and queryable knowledge layer spanning code, data, jobs, schedulers, interfaces, and business rules. That knowledge becomes the basis for deciding, slice by slice, what to re-engineer, replace, redesign, or deliberately leave unchanged.

This versioned and queryable knowledge can be exposed to IDEs and AI agents through controlled interfaces such as Model Context Protocol (MCP), allowing engineering teams to work against a trusted representation of the legacy estate rather than isolated code snippets. Combined with target-architecture principles, this guides decisions on integration patterns, batch operating models, performance guardrails, transaction boundaries, security, and runtime behaviour.

Core Reply complements this approach with a proprietary, metrics-driven framework that measures structural complexity, documentation effort, remediation effort, and quality thresholds. This enables realistic estimates of time and cost, early feasibility validation, and an incremental modernisation path grounded in evidence rather than intuition.

Core Reply’s Modernisation Framework

To ensure transformation initiatives remain feasible, governable, and aligned with corporate objectives, Core Reply’s Modernisation Framework applies an exhaustive, data-driven methodology across distinct phases.

Assessment & Discovery

The first objective is to recover knowledge. The landscape is analysed across source code, data models, batch jobs, schedulers, interfaces, and operational processes to identify dead code, hidden dependencies, business criticality, and non-functional constraints.

Strategic Blueprinting

A target high-level architecture is defined to manage coexistence between the newly modernised system and remaining legacy applications. This involves establishing explicit integration patterns, such as synchronous and asynchronous communication, event-driven architectures, Change Data Capture (CDC), and strategies for distributed transaction management.

Scoping & Feasibility

A targeted Proof of Concept (PoC) is executed to evaluate technical feasibility and accurately extrapolate the required transformation effort. Rather than relying on estimations, this phase utilises specific metrics to measure the intrinsic complexity of the legacy source code, the manual effort required by functional analysts to refine AI-generated technical documentation, and the developer effort needed to elevate AI-generated code to strict target quality standards. The final code quality is rigorously measured through static analysis tools against maintainability, reliability, security, and test coverage thresholds.

Execution & Governance

Core Reply proposes an AI-powered “Code2Doc2Code” paradigm, generating new codebase architectures derived from AI-refined functional requirements rather than direct, mechanical code-to-code translation. A central Design Authority continuously oversees the project to guarantee architectural congruence, ensuring the solution is built precisely as designed. Furthermore, a robust testing strategy is enforced, leveraging automated test case generation, dual-run executions, and parallel processing to verify strict iso-functional equivalence.

Frequently Asked Questions

What are the primary drivers for initiating a legacy modernisation programme?

Reply's experience shows that organisations usually modernise for a combination of strategic reasons.

Business Drivers
Increasing agility, reducing time-to-market, and mitigating the risk associated with shrinking legacy skills.
Cost Drivers
Reducing mainframe consumption, decommissioning obsolete infrastructure, and focusing investment where it delivers the greatest return.
Technology Drivers
Preparing systems for cloud-native operating models, DevOps, resilience, and long-term reduction of technical debt.
Knowledge Drivers
Regaining visibility over systems whose logic is often accumulated over decades and understood by only a few experts.

Once this knowledge is reconstructed, clients can make informed decisions on what to modernise, replace, redesign, or deliberately keep as is.

How does Core Reply’s Modernisation Framework handle different types of legacy artefacts?

The legacy estate typically includes core source code, batch flows, schedulers, databases, interfaces, and service contracts. Reply’s approach varies by artefact.

Core Transaction Logic
For business-critical logic, a PoC is required to evaluate transformation feasibility, functional risk, and the impact of external dependencies before industrialisation.
Batch and Scheduling Components
These artefacts are analysed not only for code conversion, but for execution windows, parallelism, restartability, reconciliation, and cut-off constraints.
APIs and External Integrations
Modernisation must account for contract evolution, orchestration, event propagation, and the implications of moving from tightly coupled transactions to distributed flows.

How is the knowledge base constructed during the reverse engineering?

What testing methodologies are applied to guarantee functional equivalence?

Ensuring behavioural continuity requires a multi-layered testing strategy.

Baseline Capture
Inputs, outputs, operational indicators, and representative workloads are recorded across batch and online flows within scope.
Test Rationalisation
Cases are classified, deduplicated, enriched with business relevance, and mapped to the identified modernisation slices.
Parallel Validation
Automated tooling supports dual-run, reconciliation, high-volume execution, and performance checks to verify iso-functional behaviour, data consistency, and operational robustness.

Engaging Core Reply Experts for Governed Transformation

Modernising a core system is not about rewriting code faster. It is about regaining control, making the right architectural choices, and progressing in sequential, evidence-based increments. Core Reply combines domain expertise, disciplined governance, and a proprietary measurement framework to help clients build realistic roadmaps, validate feasibility early, and modernise with confidence, without exposing critical operations to avoidable risk.

Core Reply

Core Reply is a company within the Reply Group that is specialized in innovating the Core Systems of Financial Institutions. We are dedicated to leading transformation projects by providing consulting, design, and implementation of innovative solutions to renew existing application environments. We support our clients in bringing innovation to sectors traditionally governed by legacy systems, modernizing Core Systems to meet future needs.