OpenAI is betting that workers across the United States are ready for autonomous AI agents, according to Thibault Sottiaux, the company’s head of core products, which include ChatGPT, Codex, and the new ChatGPT Work platform. In an interview with TechCrunch, Sottiaux said the company hit 20 million users for ChatGPT Work, signaling what he called a world that “seems to be ready” for AI that can complete entire complex tasks without heavy user oversight. The product aims to take the coding-agent power familiar to Codex users and package it for a broader white-collar audience, accessible on mobile and web through the $20 per month Plus plan.
Sottiaux explained that OpenAI’s approach is to build highly capable models and then minimize the interface between the user and the technology. He described a design philosophy of “discovery,” where the company pushes model capabilities and then leans into whatever tasks the AI turns out to handle best, such as processing large document sets, generating reports and slides, or performing deep research. That process, he said, involves learning from community feedback and iterating on the product, which is part of the company’s broader strategy of iterative deployment.
When asked about the economic incentive for OpenAI to own the application layer, Sottiaux said the company’s model is straightforward: the more utility users get, the more willing they are to pay. He compared it to the core ChatGPT subscription, where users feel the value justifies the monthly fee. He also addressed concerns about pricing and token usage, noting that OpenAI recently announced an 80% price cut on its Luna model, which he called a permanent correction that will make frontier capabilities cheaper over time. His goal, he said, is for users six months from now to be able to do the same work for less money.
On the question of whether workers are ready to trust AI with sensitive access, such as email or iMessages, Sottiaux pointed to the company’s investment in safety and alignment. He said OpenAI publishes honest benchmarks and that its models are world-class on safety topics. He also acknowledged the transition from Codex, which was built for a forgiving technical audience, to a mainstream product, describing it as an exercise in making something powerful but also safe and delightful for a much larger population.
Sottiaux framed the broader mission as bringing everyone along with the technology, not just early adopters. He noted that ChatGPT Voice has seen significant growth, and he sees a future where interaction becomes more natural, adapting to humans rather than requiring users to learn complex interfaces. The company’s role, he said, is to teach people that AI can do far more than help with writing or personal advice, and that complicated tasks can now be handled autonomously.
More AI news from TechManNews.
