17 Mar 2026
21m

OpenAI’s Michael Bolin on Codex, Harness Engineering, and the Real Future of Coding Agents

Podcast cover

Turing Post

The discussion centers on the evolving roles of AI models and the "harness" (environment) in coding, featuring Michael Bolin, lead for Open Source Codex. Bolin emphasizes the importance of the harness in enabling the agent to perform optimally, particularly regarding security and sandboxing, which are critical for coding agents like Codex. He also notes Codex's significant user growth since its launch, driven by the introduction of the Codex app, which offers a "mission control" interface for managing multiple coding conversations. Bolin highlights how coding agents are changing developer workflows by increasing throughput, enabling parallel work streams, and automating code review, and he touches on the rediscovery of best practices like documentation and test-driven development due to the influence of AI agents.

Outlines

Part 1: Core Concepts, Architecture

Part 2: Performance, Adoption, Workflow

Part 3: Context, Documentation, Autonomy

Part 4: Sandboxing, Reliability, Future

Sign in to continue reading, translating and more.

Open full episode in Podwise