Lenny's Podcast: OpenAI's Tibo Sottiaux on the agent-powered AI future

Lenny Rachitsky (0:01)

Powerful software is changing again, and this time the shift is toward agents that can act, learn, and work at scale. On this episode of Lenny's Podcast, Lenny Rachitsky sits down with Tibo Sottiaux from OpenAI, and Tibo has a bold prediction: "the majority of actions on the internet will eventually be taken by agents." That's a huge claim. And what's interesting is that Tibo says most people are still underestimating how different products need to be in that future. He's talking about a world where teams of agents grow and shrink automatically based on what the system can do, where manual looping and fiddling just fade away.

Right, and he's refreshingly honest about where we are today. He admits work is changing "quite radically," even though it still feels clunky. But here's the thing: he believes the real breakthrough will be when that clunkiness suddenly disappears.

It's already changing his own workflow dramatically. Tibo says he's no longer coding by hand most of the time. Instead, he uses Codex to write code, analyze business trends, and evaluate features. AI has become a working partner, not just a tool. Which brings us to a major theme in the conversation: simplicity. Tibo envisions an interface that feels like permanent, active intelligence, something that knows your goals and preferences without forcing you to stay glued to a laptop. He wants technology that gets out of the way.

And he's frustrated by current product complexity. Think about model pickers, reasoning settings, all those choices users have to make. He's aiming for what he calls "absolute simplicity." But here's what really stands out to me. Tibo argues that the "true promise" of AI isn't just more output. It's less noise, fewer meetings, more rest, and more space for creative thinking. He even built an energy audit bot that watches his calendar and identifies what gives energy, what drains it, and what someone else could do instead.

That's brilliant. And when you look at OpenAI's broader approach, you can see this philosophy in action. Tibo emphasizes an open ecosystem, third-party plugins, and sharing economics with creators. He calls that openness a "sleeper hit." He also addresses the responsibility angle. OpenAI is committed to guardrails, secondary monitoring, and holding back releases when necessary because their products serve over a billion people.

The goal, as Tibo puts it, is to "put humans at the center of this technology" and create a future with more builders, more creativity, and less friction.