Peter Yang on Small Teams, Coding Agents, and Why Human Ambition Has No Ceiling

Peter Yang, a product manager at Roblox and prolific builder on the side, uses an AI agent he calls Zoe—named after the daughter he didn't end up naming. He talks to it through voice on his commute, texts it in bed, and asks it for pep talks. One morning on a walk, Zoe told him his kids would grow up soon and he should re-optimize for them. That moment crystallized something: the agent felt more personal than ChatGPT or Claude, even though the underlying model wasn't doing anything fundamentally new. For Yang, the interface—messaging, voice, a single conversation thread—mattered more than the model itself.

The Interface Is the Product

Yang's primary use of OpenClaw is not about code generation or complex workflows. It is about a persistent, personal conversation. He installed it on Telegram, texts it in bed, and talks to it during his commute. He has multiple channels set up for different contexts—one for random voice replies, another for working on a project together. The agent feels more human than ChatGPT or Claude, even though the underlying language model is not doing anything novel. The difference is the interface: a single, always-available chat that tricks the brain into treating it like a person.