Thibault Sottiaux, who leads OpenAI’s core products including the API, agent infrastructure, enterprise offerings and all ChatGPT variants (ChatGPT Work and ChatGPT classic), as well as Codex, discussed the company’s strategy for bringing agent capabilities to a broader set of users.
Role and reporting
Sottiaux confirmed that the product teams ultimately report up to Greg Brockman: “I like to say that everyone reports to Greg at the end of the day.” He oversees core product areas that include Codex and the various ChatGPT products.
Goal: bring agents to a wider audience
According to Sottiaux, ChatGPT Work is intended to make capabilities that were previously aimed at technical users accessible to white‑collar workers in a safe, delightful way across mobile and web. The product was launched as part of the ChatGPT Plus plan, which costs $20 per month.
He framed the effort as making the technology available to the broadest possible population while keeping the experience secure and useful.
Product philosophy: discovery and simplicity
The team treats product development as partly a process of discovery: as they push model capabilities, they learn what the models can do and then focus on the highest‑value use cases. Sottiaux cited GPT 5.6 as an example of a step forward for general work tasks — processing large volumes of documents, generating high‑quality slides and reports, and conducting deep research — and said they iterate based on user feedback.
They aim for a minimal, natural product surface that lets the model “express itself” so users can harness its full utility without complex interfaces getting in the way.
Voice interface and natural interaction
Sottiaux mentioned the launch of ChatGPT Voice, which has seen significant growth and enables more natural, conversational interactions. He argued that the progression of the technology should make interactions increasingly human‑centric, so users don’t have to adapt to the app but the model adapts to users.
Adoption and scale
Sottiaux said the market appears ready for a more ‘magical’ level of assistance: OpenAI announced that it reached 20 million users. The product was rolled out with a focus on being simple yet powerful.
Cost efficiency and pricing
OpenAI is working to improve efficiency daily. Sottiaux highlighted a major price reduction tied to Luna — an 80% cut he described as a permanent price correction. The long‑term goal is to deliver more utility for the same dollar amount over time, so that users can accomplish the same tasks for less spend months down the line.
Data access and safety
When asked about users’ concerns over granting access to email or iMessages — functionality that was rolled out the same day as the interview — Sottiaux stressed the importance of choosing safe, aligned models. He said OpenAI invests heavily in its safety stack, publishes honest benchmarks, and that their models perform at a world‑class level on these topics.
Closing
Sottiaux framed the work on ChatGPT Work as an iterative loop between pushing model capabilities, discovering practical use cases, and refining the product based on real usage. The stated aim is to enable advanced autonomous task handling not only for technical users but for a broader white‑collar audience, while continuing to improve cost efficiency and safety controls as adoption grows.
