Skip to content

Governance Levels

The HugAI methodology defines three levels of governance for the interaction between AI agents and humans:

  1. Autonomous
  2. The AI agent can perform tasks and make decisions within a predefined range without requiring prior human approval. Used for low-risk or highly repetitive tasks, but always with traceability and the possibility of subsequent audit.
  3. Supervised
  4. The AI agent can propose actions or solutions but requires human review and approval before executing significant changes. This is the recommended mode for medium-impact tasks or when additional control is needed.
  5. Manual
  6. The AI agent can only suggest actions, but the final execution depends entirely on human intervention. Reserved for critical, sensitive tasks or those requiring expert judgment.

The choice of governance level depends on the context, the risk associated with the task, and the organization's policies. Levels can be dynamically adjusted according to the project phase or the agents' performance history.