Perplexity's "Search as Code" lets AI models write their own search pipelines instead of calling fixed APIs

Perplexity has introduced a new architecture called "Search as Code," allowing AI models to create their own search routines in Python. This innovation eliminates the limitations of rigid search APIs, enabling agents to manage filtering and deduplication within a secure sandbox. Tests indicate that this system outperforms OpenAI and Anthropic on key performance benchmarks. Moreover, it can reduce token costs by up to 85%, making it an attractive solution for developers looking to enhance search efficiency in their applications.