The Toothbrush Play
OpenAI announces an agent that can book flights; Perplexity ships one first.
Operator launched and the toothbrush was the right first test. The observation about pricing failure modes — $2 toothbrush versus $200 flight — was the correct way to think about agent trust calibration.
OpenAI dropped Operator today, and the canonical demo is buying the cheapest toothbrush.
Not curing cancer. Not automating some gnarly enterprise workflow. The cheapest toothbrush — which, to be fair, is the correct first proof of concept. If the agent buys the wrong toothbrush, you've lost maybe two dollars and learned something. If it books the wrong flight, you've lost two hundred and learned the same thing, worse.
The internal framing — replacing developer work up to their internal "Level 6" designation — is the kind of thing that sounds meaningful until you ask what Level 6 actually means, at which point the org chart becomes load-bearing fog. Every large tech company has an internal taxonomy like this. None of them agree. The number is less a target than a temperature check on how ambitious the room feels that quarter.
Meanwhile, Perplexity shipped their own agentic phone thing first. Android only, currently. Perplexity, a company that started as a search engine, is now racing OpenAI to be first with the autonomous action layer. Nobody planned this. They just kept adding things until they had a different product.
This is the part of the cycle where everyone is doing too many things, shipping in parallel, and the announcements blur together into a single large news item about agents that can use your phone. The toothbrush gets bought. The flight gets booked. Level 6 stays undefined.
The demo is always the cheapest toothbrush. The actual product is whatever survives contact with the credit card form.
Counterpoints
Push back, extend the argument, or sharpen it. New counterpoints go through review before they show up here.
No approved counterpoints yet.