The video explains how to start a project task in a free AI coding mode and contrasts it with eco and prime options. The free tier uses basic models for light work without a subscription, while the other modes use subscriber quotas or paid credits.
A prepared JavaScript shopping-cart function provides the practical test. The agent first reproduces four failing tests, then converts string prices and quantities to numbers, includes quantities in the total, rounds to cents and reruns the unchanged test suite with all five tests passing.
The result is encouraging but deliberately narrow. The video recommends giving the agent an expected outcome and a verification method, monitoring both its five-hour allowance refresh and weekly cap, and starting with focused work such as code explanation, small interface changes or documentation that can be checked independently.
Watch on YouTube
