On July 23, Tencent Cloud officially launched the new cloud-based R&D AI agent CodeBuddy NPC. As a major upgrade product of the CodeBuddy code assistant for enterprise R&D scenarios, this AI agent deeply relies on the Tencent Cloud CNB R&D platform and natively integrates Issue, code repository, PR, and CI/CD process data, achieving end-to-end self-developed R&D in the cloud.

Different from traditional on-premises AI coding tools, developers only need to issue task instructions in the ticket, and CodeBuddy NPC can independently complete the entire workflow, including solution planning, coding development, submitting PRs, and automated testing, and can self-iterate and fix errors based on CI run results.

In addition, the product supports orchestrating multi-role collaborative NPC Teams, enabling division of labor in project management, content creation, asset integration, and code reviews. In terms of computing costs, through optimizing prompts, tool calls, and cache hit rates, the official CodeBuddy NPC has reduced the first-round Token consumption by more than 90% (from over 20,000 to about 2,000), and supports flexible model switching according to task difficulty.

As AI coding tools evolve from simple code generation to intelligent agents with autonomous planning and execution capabilities, the release of CodeBuddy NPC marks that the enterprise-level AI R&D ecosystem has entered a stage of deep engineering collaboration. It significantly reduces the cost of large model applications while promoting the comprehensive intelligent reconstruction of cloud-native R&D workflows.