Perplexity Uses GPT-6 Astra for End-to-End Systems With Less Oversight
Perplexity says GPT-6 Astra can handle coding, testing, communications, and production monitoring with less frequent human check-ins.
Perplexity is using GPT-6 Astra to handle more of its software work with less frequent human oversight. In a new customer story published by OpenAI on September 14, Perplexity says Astra can work across end-to-end systems, from testing applications to changing software and monitoring production systems.
How Perplexity is using GPT-6 Astra
Perplexity says better coding models also improve its search engine because they can write programs that search the web and internal information, then summarize the results. The bigger change with GPT-6 Astra is that the model can apply those capabilities to real-world systems rather than only generating code or answers.
According to Perplexity Cofounder and Chief Strategy Officer Johnny Ho, the company uses Astra to craft communications, edit real-world systems, and monitor production software. OpenAI says Perplexity can now trust the model with full end-to-end systems and check in much less frequently than it did with earlier models.
Astra is also being used to test applications
One of the clearest examples is software testing. Perplexity asks GPT-6 Astra to build a small testing program around an application. The model can generate realistic responses that another service might return, such as a language model API or connector, then use those responses to test how the application behaves from start to finish.
That matters because the model is not simply writing a test case. It is being used to simulate parts of the surrounding system and evaluate the full workflow, which can reduce the amount of manual checking engineers need to do.
Why this matters for AI coding agents
The Perplexity example shows how AI coding systems are moving from code generation toward longer-running engineering workflows. GPT-6 Astra is being used to write code, create tests, interact with real systems, and monitor production software, according to OpenAI's account of Perplexity's use.
The customer story does not provide a specific productivity percentage or independent benchmark from Perplexity. Instead, its main claim is about a change in how much oversight the company needs while using Astra across end-to-end systems.