← Full glossary
Human-on-the-loop
gobernanzaHOTL
Human-on-the-loop is the model where the agent acts on its own and a person supervises, able to intervene or stop it, instead of approving each action as in human-in-the-loop.
Implementa formulation, drawing on the wider autonomy and decision-systems literature. In human-in-the-loop a person approves each action before it runs; in human-on-the-loop the agent executes while the person monitors, reviews by sampling or alerts, and can interrupt. The choice depends on action risk: reversible, low-impact actions allow supervision; irreversible actions involving money or personal data usually need prior approval.
How it differs from
- Human-in-the-loop
- In HITL the person approves before each action; in HOTL the person supervises and can step in during or after.
Examples
- A reconciliation agent posts small items on its own while a person reviews a daily report and the exceptions.
FAQ
- When to use HITL versus HOTL?
- Implementa rule of thumb: prior approval for irreversible actions, or those involving money or personal data; supervision for reversible, low-impact actions, with logging and an undo path.