‘The world seems to be ready’: An interview with OpenAI head of product Thibault Sottiaux
TechCrunch Tim Fernholz
OpenAI says ChatGPT Work is for everyone now, not just coders. The pitch is simple: more people, more agents, less friction — and a lot more trust needed.
Based on reporting by TechCrunch, Tim Fernholz — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI’s Thibault Sottiaux is one of the people behind Codex and ChatGPT Work, and he thinks the company is trying to do something bigger than ship another workplace tool. The goal, as he put it, is to take capabilities that started with technical users and make them safe, simple and useful enough for a much wider audience.
That’s why ChatGPT Work sits inside the Plus plan at $20 a month. Sottiaux framed that price as part of the product logic: give people enough utility that paying feels obvious. Not cheap, exactly. Just inevitable enough that the subscription stops feeling like a tax and starts feeling like a bargain.
He also kept returning to the same theme: the model should lead, and the interface should get out of the way. OpenAI, in his telling, is moving from text to voice to more autonomous tasks because the technology is becoming more natural to use. People shouldn’t have to learn the app, he said. The app should adapt to them.
That “magic” approach is being pushed further with work tasks. Sottiaux said ChatGPT Work is now being used by 20 million users, and described the product as a place where the company can discover what the models are actually good at. He pointed to GPT 5.6 as a step up for processing lots of documents, making slides and reports, and doing deep research — the sort of chores professionals already spend time on.
There’s a catch, of course. If OpenAI wants workers to hand over email and iMessages, the company has to earn the trust it keeps asking for. Sottiaux said safety and alignment are central to that effort, along with honest benchmarks. OpenAI is betting that people will accept more autonomy if the results are useful enough and the guardrails are visible enough. That’s a familiar Silicon Valley move. The only difference is that this time the thing being asked to earn trust is not a social app or a scooter, but a software agent with your inbox.
My take — AI-written commentary, not fact-checked reporting
This is the usual OpenAI wager: ship the magic first, explain the boundaries later, and trust that “safe and delightful” will cover the gap. It’s a clean story until the product wants your email, your messages, and your patience. The broader pattern is clear: the industry keeps calling autonomy a UX improvement right up until somebody asks who’s liable when the robot gets creative.
Read more about this at: TechCrunch
