Five minutes to find out whether autonomy has moved faster than accountability.
Ten questions. One practical test: can you state what your AI agents are allowed to decide, what they are allowed to do, and who remains accountable when they do it?
Score what you can prove, not what you think is probably true.
For each question choose 2 = Yes, clearly and evidenced, 1 = Partly / depends, or 0 = No / not sure.
Red-flag override: a score of 0 on consequence limits, delegation, stop authority or accountability should be treated as a material boundary concern regardless of the total score.
What does the boundary look like?
Complete the test to see your boundary position.
Your result will appear here once all ten questions are answered.
Can you state the delegation in one sentence?
If this sentence is difficult to complete, that is probably the finding.
The Quick Test finds the question. The full Check tests the evidence.
The full Agentic AI Decision Boundary Check™ examines 25 evidence-based criteria across Purpose, Authority, Constraint, Intervention and Accountability. It tests not only what management believes the agent can do, but what permissions, tools, controls, logs and operating evidence show it can actually do.