How much autonomy have your agents earned?

About two minutes. See where your team stands between the autonomy you give your agents and the autonomy you can account for.

Your answers stay in your browser until you choose to send them with your email at the end.

Roughly how much of your code do agents write?

Under 10% 10–40% 40–70% Over 70%

How do your coding agents usually run?

Suggestions only A person approves each action Some actions are pre-approved Mostly in auto mode

Who can approve an agent-written change for merge?

(Choose the one that goes furthest.)

Only a person A person, or a rule for some low-risk changes An AI reviewer can approve No approval is needed for some changes

Do agents act beyond code: deploys, infrastructure, data, tickets?

No Read-only Some write actions Broadly

If someone asked who allowed a specific agent action last week, could you show them?

No Only by digging through logs Quickly, for some of our tools Quickly, for all of them

Can you tell whether an approval came from a person, a rule you wrote, or an AI reviewer?

No Sometimes Mostly Always

How do you know a review meant something, and wasn't a rubber stamp?

We don't Occasional spot checks We track review signals, like time to approve or comments We regularly check reviews against a careful second look

Could you hand an auditor or customer evidence about your agent-written changes today?

No We'd assemble it by hand Partly automated Yes, ready to export

Do you hold SOC 2 or ISO 27001?

Which compliance platform do you use, if any?

When is your next audit?

Do customers' security reviews ask about AI-written code?

Often Sometimes Not yet Don't know

Optional

Do you expect to give agents more autonomy in the next six months?

Yes Maybe No

Optional

What holds you back from giving agents more autonomy?

(Choose any.)