-
Notifications
You must be signed in to change notification settings - Fork 0
the human agent orchestrator
Nicolas Cravino edited this page Jul 29, 2026
·
1 revision
id: the-human-agent-orchestrator title: The Human-Agent Orchestrator tags: [llm] category: agentic-frameworks published: 2026-07-29 created: 2026-07-29 updated: 2026-07-29 freshness: fresh date_source: git-commit source_file: The-Human-Agent-Orchestrator.md
On the book The Human-Agent Orchestrator and the contributed Orchestration Design Canvas: six layers — Source, Success, Safety, Steering, Switch, Sharpen — that turn "deploy an agent" into something a board can defend.
- Most agent failures come not from the model but from the orchestration layer — the part leaders skip and operations teams eventually inherit. [^src:articles-published §the-human-agent-orchestrator]
- The Canvas structures the hard questions: source of truth, definition of success, safety enforcement, steering, when control switches to a human, and how the workflow is sharpened over time. [^src:articles-published §the-human-agent-orchestrator]
- Agent projects fail less because the model is weak and more because the operating design is underdeveloped; the canvas makes those gaps visible. [^src:articles-published §the-human-agent-orchestrator]