AI development drives increase in untested code deployment in Japan

The rapid integration of generative AI into software development is creating a gap between code production speed and quality assurance capabilities. According to the Tricentis ‘Quality Transformation Report 2026’, 65.6% of Japanese companies are deploying software containing untested code into production environments. This trend is driven by the pressure to increase release speeds and the overwhelming volume of code generated by AI, which exceeds the capacity of current testing teams to verify.

While AI improves productivity by automating coding and documentation, it also introduces risks such as security vulnerabilities and system inconsistencies. To mitigate these issues, industry insights suggest that Data Streaming Platforms (DSP) are becoming essential. A survey of 4,625 IT leaders indicates that 88% prioritize DSP investment to provide the real-time, high-quality data necessary to reduce AI hallucinations and support agentic AI. Implementing ‘shift-left’ processing—handling data closer to the source—is also highlighted as a key method to ensure data quality and reduce processing costs.