Sixteen Claude AI agents working together created a new C compiler (arstechnica.com)

Anthropic has showcased a remarkable achievement in AI coding by having 16 instances of its Claude Opus 4.6 model collaboratively create a new C compiler with minimal supervision. Over two weeks and approximately $20,000 in API fees, these AI agents were tasked with developing a compiler from scratch, resulting in a 100,000-line Rust-based compiler capable of building a bootable Linux kernel across multiple architectures, including x86, ARM, and RISC-V. This project utilized a novel "agent teams" feature, allowing each Claude instance to operate independently within its own Docker container, effectively managing tasks and resolving conflicts without a central orchestrator. This accomplishment is significant for the AI/ML community as it demonstrates the potential for collaborative, semi-autonomous AI systems in complex coding tasks, paving the way for more advanced developments in software engineering. The resulting compiler, which is available on GitHub, boasts a 99% success rate on a rigorous testing suite and has even successfully run classic software like Doom. However, while the project highlights AI's capabilities in defined environments, Carlini notes that real-world software development poses greater challenges due to the ambiguity of tasks and the dynamic nature of requirements.