NTT Launches Innovative Technology for AI-Optimized Software Assets

In a significant advancement for software development, NTT, Inc. has introduced innovative technology aimed at transforming legacy software assets into formats that generative AI can easily comprehend and utilize. This move addresses longstanding challenges in the software industry where documentation, code, and testing often fall out of sync, particularly in well-established systems.

Challenges in Utilization of Generative AI

Many organizations have recognized the potential of generative AI to improve software development efficiency, yet the focus has predominantly been on new projects. With about 60% of software work centered on updating, maintaining, or redeveloping existing systems, enhancing productivity in this area is crucial. Developers of older systems frequently grapple with inconsistencies between documentation and actual code, making it tough for AI to function effectively in these environments.

Moreover, when existing systems exhibit outdated or inconsistent documentation, the complexity increases for developers trying to identify necessary changes and the possible impact of those changes. This disparity reduces the effectiveness of generative AI, exacerbating the struggle to validate the integrity of modifications, especially if existing tests are inadequate.

Innovative Features of NTT's Technology

NTT's new technology employs two key components designed to enhance the AI-readiness of legacy software assets:

• Traceability Analysis: This technology organizes and interprets the relationships between specifications, design documents, code, and testing artifacts. It adeptly identifies mismatches and deficiencies across different development stages.
• Automated Test Generation: This feature pinpoints perspectives not previously covered by tests. It generates additional tests to address these oversights systematically.

Together, these advancements allow AI to thoroughly understand the structural intricacies of existing systems and facilitate a proactive approach to modifications, leading to a smoother transition towards AI-Native development methodologies.

Preliminary Successes

In initial trials of the new technologies, NTT evaluated their ability to detect errors embedded within design documents. Notably, the new system achieved an impressive 80% detection rate for inconsistencies—marking a drastic improvement compared to the conventional method, which only identified 27% of errors. Additionally, the new technology not only highlighted existing discrepancies but also unearthed previously unknown latent bugs, emphasizing its proficiency in enhancing software quality.

The trials utilized a sample application, the Macchinetta Framework, which served to showcase the capabilities of NTT's advancements in practical settings. The methodologies underlying the technology ensure a comprehensive depiction of dependencies, enabling effective verification of impacts stemming from changes made to existing functions.

Looking Ahead

NTT is committed to refining this technology further by applying it across various existing systems, expanding its compatibility with diverse programming languages, and honing the mechanisms for capturing tacit knowledge. There are plans to broaden the quality metrics and testing techniques supported by this technology, with real-world testing on ongoing software projects set to validate its practical application fully.

By streamlining the transition to AI-Native development approaches, NTT aims to assist organizations in maintaining their current systems in an accessible, AI-friendly state. This innovative approach promises to reshape the development landscape, enhancing both productivity and quality in software maintenance.

"Our goal is to make AI integration seamless across all stages of software development, from conception to execution, ensuring organizations can fully leverage existing systems with ease and precision," said an NTT spokesperson.