If Agents Must Choose What to Do Next, Why Is the Instructor Hard-Coding the Entire Flow?

[therisingsun]

My question is simple - In Module 1, the instructor repeatedly emphasizes that the ability to decide what to do next—i.e., choosing among options—is the key factor that separates an agentic system from a non-agentic one.

Yet in multiple examples, the instructor himself hard-codes the entire workflow (planner → researcher → fact-checker → writer). The developer is wiring up the flow explicitly, leaving no actual decision-making for the agents.

So how is this agentic at all?
Isn’t this contradicting the core principle being taught?
I don’t understand why the instructor keeps calling this an agentic system when the execution sequence is fully predetermined by the him(the developer.). Even if it is not sequential like 1,2,3,4. Lets assume its some other flow say repeat(1,3,2), then (2,4) or 10 other patterns. The fact that the instructor (developer) is hard coding the pattern makes it non-agentic and contradictory to what he’s emphasizing ! Please reconcile So how is this agentic at all?Isn’t this contradicting the core principle being taught? Thanks

[lukmanaj]

Hi, welcome to the community.

Yours is a very common and valid question, because it does look like a contradiction on the surface.
The key is that "agentic” does not necessarily mean “the agent chooses the high-level pipeline.”
It means something more specific:

An agent is agentic if, within its scope of responsibility, it can decide what to do next.

In essence, all the agents in the crew are still agentic, they will work on tasks related to them by deciding on the tools to use and actions to take before passing on to other agents. Hope this clarifies it.