Quiz: Can Your AI Agent Recommend, Prepare, Approve or Execute?
A quick scenario quiz for defining how far an AI agent can go before human approval, escalation or accountability is required.
Would You Let This Agent Recommend, Prepare, Approve or Execute?
The agent can draft the order but can it send it? Approve it? Pay it? Somewhere between “helpful assistant” and “please explain this to Finance” lives the delegation boundary.
Take five quick questions to see whether your autonomy settings have actual decision boundaries or just optimistic verbs. “The agent can do it” is not the same as “the agent is allowed to do it.”
How Far Down the Workflow Can Your Agent Walk Alone?
Five quick questions but “Execute” is not “Prepare” with better permissions.
Match each level to the behaviour.
Word bank: Recommend · Prepare · Approve · Execute
If an agent is allowed to draft a payment instruction, it is automatically allowed to send the payment.
An AI agent finds a supplier invoice that matches the PO and delivery record, what can the agent do do?
Choose the line that best describes your delegation model.
A case falls outside its normal rules. What happens?
Does this feel familiar?
Try the free five-minute check: What Is This Agent Actually Allowed to Do?
Want the deeper version?
Deeper Insight and the matching toolkit are future subscriber drops. Subscribe for the next Pattern release.
Capability says can. Delegation says may.
Pattern spotting, not an organisational diagnosis. The useful question is whether the behaviour, owner, cost or learning route is visible.
So… what did you just spot?
The core distinction is simple: clear delegation levels separate assistance from action and connect agent autonomy to human accountability.