Inference by Turing Post OpenAI’s Michael Bolin: What Engineers Still Matter For in the Age of Coding Agents
6 snips
Mar 24, 2026 Michael Bolin, tech lead for Codex at OpenAI who applies large language models to developer tools, discusses what programming looks like when agents write most code. He explores shifting mindsets to build for agents, which skills will matter, faster prototyping, risks of losing human judgment, and where human taste and oversight remain essential.
AI Snips
Chapters
Books
Transcript
Episode notes
Programmer Identity Evolves Not Disappears
- Michael Bolin says programming identity shifts as agents write code but the goal remains building useful artifacts.
- He still shapes outcomes and applies personal experience even when not physically typing code, calling it a different workflow.
Agents Make Prototyping Orders Of Magnitude Faster
- Michael Bolin reports he builds more now because prototypes are cheap with agents.
- Tasks that once took a weekend now execute quickly, enabling far more experimentation and iteration.
Ask Better Questions Not Just Write More Code
- Learn to ask the right questions and be curious because you're still in charge of shaping code produced by agents.
- Be comfortable reading code, tracing abstractions, and interrogating why a design or implementation exists.
