Agentic coding at ClickHouse

In the evolving landscape of AI-assisted coding, ClickHouse has embraced the use of coding agents to enhance productivity and streamline various tasks, from writing boilerplate code and finishing stale pull requests to investigating complex bugs and optimizing performance. The article discusses the nuanced perspectives on AI usage, highlighting both the skepticism and the transformative potential of coding agents, particularly with advancements seen in tools like Claude Opus 4.5. ClickHouse has integrated these agents not only for mundane tasks but also for complex problem-solving, demonstrating significant improvements in efficiency and code quality. The company’s approach emphasizes using AI as a complementary tool rather than a replacement for human engineers, advocating for cautious and thoughtful integration to avoid pitfalls such as AI psychosis and maintaining a focus on core product strategies. With AI's potential to reshape engineering practices, ClickHouse encourages experimentation while maintaining rigorous standards to ensure that AI augments rather than detracts from human expertise.