Any patterns with human-in-the-loop? I am referring to the Partial to Full Autonomy journey that Karpathy talked about, which of these patterns would fit that?
Good question. In general the "AI agent design patterns" I've mentioned don't mention the human-in-the-loop explicitly. In some cases its implicit, like when you approve the research plan or coding plan in Deep Research or AI coding assistant. Generally, its a checkpoint in the sequential chain where the human would correct or revise.
Your section on the orchestrator-workers workflow is spot on. I used exactly that pattern to build a multi-agent code review system. One lead agent analyses the diff, dispatches frontend/backend/devops specialists only for the files that need them. Anthropic's own docs got me thinking about it this way: https://reading.sh/one-reviewer-three-lenses-building-a-multi-agent-code-review-system-with-opencode-21ceb28dde10
Any patterns with human-in-the-loop? I am referring to the Partial to Full Autonomy journey that Karpathy talked about, which of these patterns would fit that?
Good question. In general the "AI agent design patterns" I've mentioned don't mention the human-in-the-loop explicitly. In some cases its implicit, like when you approve the research plan or coding plan in Deep Research or AI coding assistant. Generally, its a checkpoint in the sequential chain where the human would correct or revise.