AWS Transform Custom Tackles Technical Debt

AWS recently announced AWS Transform Custom, a capability within AWS Transform that allows developers modernize and refactor custom code.

The AI-driven service provides out-of-the-box transformations for common scenarios, such as upgrades to Java, Node.js, and Python. Furthermore, it also performs custom, organization-specific transformations, such as version upgrades, runtime migrations, complex language translations, and architectural changes.

By continually learning from code samples, documentation, and developer feedback, the agent delivers, according to the company,  high-quality, repeatable transformations without specialized automation expertise, enabling organizations to scale their modernization initiatives effectively.

In an AWS news item, the authors write:

For a typical organization, AWS Transform custom can scale modernization across hundreds or thousands of applications, achieving transformation up to 5x faster than when done manually. The transformation agent automatically captures feedback and continues to improve over time, so each subsequent transformation becomes more reliable and efficient.

(Source: AWS News Blog)

AWS Transform custom offers both CLI and web interfaces for different modernization needs. The CLI supports natural-language transformations on local codebases, suitable for both interactive and automated use. Furthermore, it can also integrate into modernization pipelines. At the same time, the web interface focuses on campaign management, enabling teams to track and coordinate transformation progress across multiple repositories.

In a LinkedIn post, Software Engineer Michael Fowlie wondered why developers would use this over Codex CLI or countless other AI coding tools. Jas Chhabra, Sr. Manager, Head of Engineering for GenAI/LLM/ML Agentic services at AWS, provided a detailed response:

It’s not a single developer focused coding tool. It’s an enterprise focused tool where central team can create and run repeatable modernization tasks using organization specific knowledge and policies. It also automatically learns your organization’s coding knowledge to get better over time. Having said that even for single transformations, you will see the difference in completeness and quality.

The announcement of the AI-driven service targeting the costly problem of technical debt sparked a discussion on a Reddit thread about the real-world accuracy and cost of automated refactoring. One respondent commented:

The entire point of these tools is that the cost to refactor is essentially zero. That’s the point of the tool.

However, others expressed deep skepticism that an AI can truly handle the institutional knowledge embedded in old applications, predicting significant cleanup work will still be required:

AI falls flat when needing to account for all the hidden business logic in legacy apps, and that’s the biggest reason folks maintain them instead of rewriting them.

Lastly, more details are available on the documentation pages.