Charlie Guo

Guo says he began Artificial Ignorance to understand developments in AI by reading, experimenting and writing regularly. In January 2026, he announced that he had joined OpenAI’s developer experience team. He described the role as helping people learn and build through documentation, examples and tutorials, across a growing surface that includes the API, Codex and ChatGPT apps.[1]

His OpenAI guide combines three parts of a long-running agent workflow: skills hold reusable procedures, a shell supplies an environment for real computation, and compaction preserves continuity as the conversation grows. The article explains how a model selects a skill from its metadata, reads the detailed procedure and executes work in a local or hosted terminal.[2]

The guide recommends workflows that install dependencies, retrieve information and write a concrete artifact. It also treats skill descriptions as routing instructions: they should explain when a procedure applies and what success produces. These details matter because an agent needs both a way to choose the right workflow and a deliverable that a person or application can inspect afterward.[2]