Veriserve Establishes Guidelines to Standardize Generative AI Quality Assurance Services

Veriserve, a provider of software quality improvement support services, has established internal guidelines for generative AI quality assurance services. By standardizing processes for AI-integrated products and AI-driven development, the company aims to enhance the skill sets of its QA personnel, ultimately increasing the quality and speed of its service offerings.

Background: The use of generative AI in software development has become commonplace. However, generative AI products possess characteristics where output tendencies change based on models or usage environments, and responses can vary under the same conditions, making it difficult to guarantee quality through conventional testing alone. Continuous quantitative evaluation and monitoring-improvement cycles are essential. Similarly, AI-driven development faces challenges regarding reproducibility and output consistency, for which systematic frameworks are still emerging. Consequently, Veriserve has developed and internally released these guidelines to standardize its firm-wide quality assurance processes.

Overview: The company has released the 'QA4AI Process Guideline' and the 'QA4AIDD Process Guideline,' developed by its expert consultants and R&D department. The QA4AI (Quality Assurance for Artificial Intelligence) guidelines define approaches for product-level quality assurance, detailing activities, tasks, inputs, and outputs. The QA4AIDD (Quality Assurance for AI Driven Development) guidelines define quality assurance approaches for process-level quality, particularly for coding agents, specifying activities, tasks, inputs, and outputs.

Future Initiatives: Both the QA4AI and QA4AIDD guidelines will be continuously revised in line with technological progress. Building on its track record as a leader in quality assurance, Veriserve continues to leverage 'People x Technology x AI' to create new value and contribute to a sustainable society.