Perplexity leans on GPT‑6 Astra for end-to-end systems

Perplexity's use of GPT‑6 Astra grew out of a problem the company already cared about: search quality. Cofounder and Chief Strategy Officer Johnny Ho notes that when the underlying model writes code better, the search engine improves too, because it can write better programs to search the web and internal information and summarize it concisely. The harder step, in his account, was moving from handling information to acting on real-world systems.

Astra's role in that step shows up most concretely in testing. With limited time to test manually, Ho asks the model to build a small testing program around an application. The model generates realistic responses like those another service would send, such as a language model API or a connector, and by standing in for those services it can check how the application responds and test the workflow from start to finish. Perplexity lists API among its products.

What makes Ho's comments notable is the supervision they imply. He says Perplexity trusts the model with full end-to-end systems and checks in much less frequently than with previous generations. That is a claim from a startup cofounder about his own deployment, not a benchmark result, so the stakes for other firms hinge on whether the same hands-off reliability holds outside Perplexity's workflows.