Module 01 Summary
What this module established
An agent is a system whose next step is decided during the run. That one property - not model calls, not tools, not a framework - is what forces everything else in this course: stopping rules, permissions, gates, and a way to see what it did.
Carry forward
- Three of five common workflows are scripts. A fixed sequence needs none of the machinery an agent does, and cannot loop.
- Autonomy is a dial. What grows with it is the window in which a wrong action goes unnoticed.
- A demo shows capability, not robustness. Ask what happens when a tool call fails, and whether that input was chosen.
Before moving on
Move on when you can apply the run-time-sequence test to a real workflow and defend the structure you chose, including choosing 'a script'.
