OpenAI's product chief: 'The world seems to be ready' for AI agents

Thibault Sottiaux leads OpenAI’s “core products” group — the API, agent infrastructure, enterprise, all of ChatGPT (including ChatGPT Work and classic ChatGPT), and Codex — reporting to OpenAI president Greg Brockman. In an interview with TechCrunch about ChatGPT Work, the company’s $20-a-month agentic tool built for white-collar workers, Sottiaux framed the launch as part of OpenAI’s mission to “bring everyone along”: “We wanted to bring the power of coding agents to everyone… making it available to as broad of a population as possible,” he said, pointing to the tool’s inclusion in the low-cost Plus plan. OpenAI has said Work and Codex together “just announced” reaching 20 million users.

Asked whether workers are ready for a product that, unlike Claude Cowork’s more interactive, choice-driven style, aims to act with minimal back-and-forth, Sottiaux said: “We definitely see that the world seems to be ready. This is why we’ve had incredible adoption.” He described the product’s development as “a product of discovery” — pushing model capability forward and then leaning into what the model turns out to be good at — pointing to GPT-5.6 as “a step up in general work: being able to process a large amount of documents, generating quality slides, generating quality reports, doing deep research.”

On the gap between subscription price and token cost, Sottiaux pointed to a recent 80% price cut on OpenAI’s Luna model as a “permanent price correction” rather than a temporary promotion: “If you want to do more, of course you can pay more to do more, but… you wake up six months from now, you should be able to do all of the same with less spend.” Asked about giving the assistant access to email and iMessage, he said safety investment was central to OpenAI’s approach: “It’s important to pick models that are safe and aligned… our models are world-class at these topics.”

Why it matters

Sottiaux’s “discovery” framing — building capability first and finding the product around it afterward — is a bet that a more autonomous, less hand-holding agent design will win out over rivals’ more interactive approach, a wager OpenAI is already testing with tens of millions of Work and Codex users.