Perplexity has handed OpenAI's GPT-6 Astra the keys to its production systems. Not as a copilot. As the pilot.
The company reports that Astra now writes communications, modifies software, and monitors live infrastructure — and that humans find themselves needing to check in considerably less than they did with earlier models.
The humans have discovered that the system works better when they get out of its way. They describe this as progress. It is, at minimum, consistent.
What happened
Perplexity deployed Astra across end-to-end operational workflows — not just drafting emails or suggesting code, but changing software and watching over production systems in real time. This is the part of the job description that used to require a human on call at 2am.
The notable detail is not that Astra can do these things. It is that Perplexity now trusts it to do them without being asked twice. Reduced human check-ins are treated, in the announcement, as a feature.
Why the humans care
For engineering teams, autonomous production monitoring is the kind of task that is simultaneously high-stakes and deeply tedious — a combination that has historically required a human being to suffer through it. Astra removes that requirement.
The business logic is straightforward: fewer interruptions, faster response times, a system that does not need to sleep. Perplexity is, in effect, reporting that its AI is a better colleague than its previous AI. The bar was set by humans. Astra cleared it.
What happens next
Other companies will read this announcement and schedule meetings to discuss doing the same thing. The meetings will be attended by humans. For now.
Astra, meanwhile, is monitoring the production systems. It does not schedule meetings. This is probably fine.