The Archaeologist’s Copilot

2026-07-16 · Source: Martin Fowler · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure · Depth: Intermediate, extended

Summary

This article details the modernization of a 20-year-old Java 1.5 "Big Ball of Mud" codebase, originally built with Ant, which no longer compiled on modern systems. Initial attempts using LLMs with "Tourist Prompts" yielded hallucinated solutions, such as a `build.gradle` for Java 1.8 and incorrect dependencies like `commons-pool2 (v2.x)` instead of `org.apache.commons.pool (v1.x)`. Success was achieved by adopting an "Archaeologist Prompt" to ground AI analysis in evidence, leading to a "critical rewrite recommended" verdict. The process involved containing the legacy code in a Docker "Time Capsule" (Java 6/Ant on x86), lifting it to Java 8/Gradle 7.6 on Apple Silicon, hardening tests to expose silent failures, and using an AI-compiler feedback loop to systematically refactor raw types to generics, ultimately transforming it into a modern, thread-safe, and containerized library.

Key takeaway

For a DevOps or AI Engineer tasked with modernizing a legacy Java application, avoid "Tourist Prompts" to LLMs that generate optimistic, hallucinated solutions. Instead, adopt an "Archaeologist" mindset, using AI to perform forensic code audits and guide containment within a Docker "Time Capsule" before gradual, test-protected refactoring. This approach ensures verifiable baselines and prevents blind changes to fragile systems, transforming opaque legacy code into manageable technical debt.

Key insights

AI is most effective in legacy code modernization when grounded in evidence and constrained by clear roles and a step-by-step strategy.

Principles

• AI defaults to optimism; restoration requires critical analysis.
• Containment precedes repair for fragile legacy systems.
• Momentum is oxygen; avoid fighting tooling.

Method

Modernize legacy Java by first auditing with an "Archaeologist Prompt," then containing in a Docker "Time Capsule" (Java 6/Ant on x86), lifting to Java 8/Gradle 7.6 on ARM64, hardening tests, and using an AI-compiler feedback loop for refactoring.

In practice

• Use Docker Compose to emulate legacy network/filesystem paths.
• Enable `-Xlint:unchecked` for targeted AI refactoring.
• Migrate `main()` scripts to JUnit 5 for granular reporting.

Topics

• Legacy Code Modernization
• AI-Assisted Development
• Docker Containerization
• Java 1.5
• Gradle Build Tool
• Brownfield Projects

Code references

• nikmalykhin/java-blobstore

Best for: Software Engineer, DevOps Engineer, AI Engineer

Related on AIssential

• Reimagining API modernization with deterministic AI-assisted engineering — AI-assisted, instruction-driven frameworks accelerate legacy API modernization by codifying patterns and ensuring deterministic outcomes.
• ScarfBench: Benchmarking AI Agents for Enterprise Java Framework Migration — AI agents struggle with enterprise Java framework migration, achieving low behavioral success due to complex dependency management.
• The New Shape of Amnesia: Technical Debt, Cognitive Debt, and the World Models Our Agents Refuse to… — AI-assisted development creates "cognitive debt" by eroding understanding of code's rationale, demanding persistent world models.
• Can coding agents relicense open source through a “clean room” implementation of code? — AI-assisted code rewrites challenge traditional "clean room" definitions and open-source licensing compliance.
• I Built an AI That Modernizes Legacy COBOL — And Then Made It Prove Its Own Work — Separating correctness (compiler theory) from intent (LLM semantics) is key to reliable legacy code modernization.
• Why AI is raising the stakes of the software engineering craft — AI elevates software craftsmanship by shifting focus to human orchestration, auditing, and deep system understanding to prevent cognitive debt.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Martin Fowler.