Lenny's PodcastOpenAI’s Head of ChatGPT: We’re entering a new era of AI (again) | Tibo Sottiaux
At a glance
WHAT IT’S REALLY ABOUT
ChatGPT’s future: persistent agents, simpler UX, and agent-driven internet
- Tibo Sottiaux lays out a vision where persistent, always-on AI agents ("Dots") learn user goals and preferences and operate seamlessly across devices, apps, and modalities.
- He predicts that agents will take the majority of actions on the internet, creating new product requirements around scale, agent-facing interfaces, and economics as agent-driven traffic surges.
- OpenAI’s product direction aims to reduce user-facing complexity by merging experiences (Codex and ChatGPT, Work and consumer) and eliminating model pickers in favor of intent-based interaction.
- He highlights an “open ecosystem” strategy—partner sign-in, plugins, discovery based on retention/quality, and revenue sharing—as a major underappreciated lever for adoption and developer opportunity.
- The conversation addresses human impacts (fatigue, loneliness, pressure) and OpenAI’s approach to safety through strong guardrails, monitoring compute, and cautious release decisions.
IDEAS WORTH REMEMBERING
5 ideasAgents should learn continuously—users shouldn’t hand-wire workflows.
Sottiaux argues that today’s tooling still forces humans to manage agent “loops/graphs,” but the end state is an always-on assistant that learns your goals and preferences from feedback and behaves proactively across contexts.
Design products for agent traffic and economics, not just human UI.
He expects agents to generate most internet actions and traffic, which will force companies to support agent-friendly interfaces (e.g., MCP-style endpoints), withstand new load patterns, and rethink unit economics when agents drive usage spikes.
The UX trend is toward fewer toggles and invisible model selection.
He describes a consolidation path: merging ChatGPT “Work” and consumer experiences, bringing Codex capabilities into ChatGPT, and removing model pickers so users simply state intent while the system chooses the right model and tools.
The “always-available, everywhere” agent replaces app-centric computing.
Dots are positioned as persistent assistants available across any client/screen—joining meetings, taking notes, continuing work via email/text—reducing the need to be “glued” to a laptop and lowering configuration overhead.
Plugin success will be driven by real retention—and monetized by revenue share.
OpenAI’s plugin ecosystem emphasizes discovery via observed utility (retention, quality, task success) and adds shared economics so popular plugins get paid via subscriber usage routed through plugins or “Sign in with ChatGPT.”
WORDS WORTH SAVING
5 quotesThe majority of actions on the internet will be taken by agents.
— Tibo Sottiaux
Over time, you just want a system that learns from, you know, what you want to achieve. You don't want to necessarily think about, "Oh, I was just thinking I've got to loop it exactly this way in order to get results."
— Tibo Sottiaux
It's just like it's available when you need it, and it also, like, gets out of the way when you don't need it.
— Tibo Sottiaux
It still feels, like, a bit clunky, and I think it will feel clunky until it isn't, and then you're just like, "Wow," you know?
— Tibo Sottiaux
I myself even get fatigued with the model picker and the reasoning efforts and, you know, whether to use multi-agent or Ultra or, you know, what it even does, and you kind of need a PhD in model pickers, right?
— Tibo Sottiaux
High quality AI-generated summary created from speaker-labeled transcript.
