Talentica Software Unfurls Managed AI Service to Optimize Software Delivery

Talentica Software has introduced a managed software delivery service that utilizes artificial intelligence (AI) to assist DevOps teams in deploying AI-generated applications at scale. The company's CTO, Manjusha Madabushi, explained that the DevX AI Pods service employs a series of AI agents trained to understand the context required for building and deploying AI-powered applications.

These agents analyze various artifacts, including product requirement documents, codebase, test cases, and software architecture, to ensure the AI-generated application can run in a production environment. A unique evaluation framework, Correctness, Consistency, Completeness and Relevance (CCCR), assesses the generated code. This framework, developed by Talentica, undergoes validation by over 600 software engineers to ensure it meets the original criteria.

This approach allows AI agents to consider changes holistically rather than implementing them independently. The goal is not merely to generate code faster but to enable DevOps teams to reliably deploy higher-quality applications more rapidly. Madabushi pointed out that while AI coding tools can generate more code, they do not necessarily increase the rate of application deployment.

Instead, they often create excessive duplicate code, leading to bloat and potential performance issues. Consequently, the generated code can increase the technical debt that DevOps teams must eventually address. It remains unclear whether the rise of AI has led to a faster deployment rate, but there is undoubtedly an abundance of generated code.

Recent surveys indicate that over half of organizations (54%) now use AI across more than half of their software development lifecycle, with 40% reporting AI generates the majority of production code merged in the last 90 days. Within the next three years, 58% of respondents expect AI to build 80% or more of their software. However, three-quarters of respondents have encountered production issues attributable to AI, with 42% experiencing multiple incidents.

The challenge lies in determining how much of this AI-generated code is deployed and whether it has been validated before integration. Each DevOps team must decide how best to manage this increased code volume. Ultimately, achieving efficient code management in the AI era will likely require greater reliance on AI to resolve the coding issues created by AI.