Perplexity, an AI-powered search engine, has adopted OpenAI's 'GPT-6 Astra' to perform tasks such as software modifications and operational system monitoring. Johnny Ho, co-founder and Chief Strategy Officer of Perplexity, stated that as the model's coding capabilities improve, the performance of the search engine—which searches the web and internal information to provide concise summaries—also improves.

Automation of Code Testing

Johnny Ho identified code testing as one of the most useful areas for AI application. In situations where time for manual testing is limited, GPT-6 Astra is tasked with creating small-scale test programs for applications. The model can generate realistic responses that other services, such as language model APIs or connectors, might send, allowing the team to check application responses and test end-to-end workflows.

Improvement in System Autonomy

Unlike previous generation models, Perplexity is now able to trust the entire end-to-end system through GPT-6 Astra, allowing them to check results far less frequently than before. The model is expanding its practical application, handling everything from drafting communications to actual system editing and production software monitoring.