Control
Let the Human Set the Rules
Control
Let the Human Set the Rules
What it means
Humans guide how agents operate by setting boundaries, preferences, and intent. Rather than making assumptions, agents respond to this direction with flexibility. Control doesn't mean limiting autonomy, it means aligning it with human goals.
Why this matters
Related Patterns
Users define operational limits for AI behavior. The agent operates within these boundaries, avoiding unintended actions.
Instruction mode
Users define interaction boundaries by selecting input modes, guiding the agent to operate safely within intended, user-controlled scopes.
Task specific boundaries
Specific checkboxes define what the AI is allowed to change (e.g., title edits) and what it must avoid. These helps in defining clear behavioral constraints.
Monitoring
Agent only observes, all decisions are manually approved. This is useful for audit, analysis, or low-trust mode.
Guided
Agent suggests actions and it can analyze and simulate rerouting or bandwidth decisions. Very similar to co-pilot.
Full control
Agent manages the system autonomously and has full control to make decisions.
Immediate agent shutdown
A prominent “Disable Agent” toggle gives users a fast, irreversible way to halt all agent activity. This supports emergency intervention and restores user authority instantly.
Visible control settings
The system shows settings upfront allowing the user to assess whether to make decisions based on risks and live situations.
How to implement
Common pitfalls
Over-reliance on implicit controls
Assuming users understand what the agent is doing without clear communication
Boundary creep
Agents gradually take on more than intended without safeguards or oversight
Ambiguous authority
It is unclear if the agent or the human is responsible for the task, especially in case of failure states
Hidden behavior
The agent performs actions without surfacing intent or outcomes to the user