According to Nate B. Jones, a growing number of knowledge workers are making a shift: moving beyond chat-based AI assistants to agent-based systems like Codex that handle entire tasks autonomously. While only one in sixteen hundred people currently use Codex, the underlying change is significant. Rather than treating AI as an interactive chatbot, users are delegating real computer work—running code, writing messages, editing files—directly to agents. Jones argues this fundamentally changes the unit of work. Instead of a single prompt, you're establishing goals, oversight, and clear boundaries for an agent that executes across multiple steps. The critical safeguard, he emphasizes, is maintaining what he calls receipts—clear visibility into what the agent actually did before you scale up. For operators and builders, this represents an emerging shift in computer literacy: less about prompting the right question, more about directing an agent and verifying its work.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton