Workflow orchestration across agents in Agentic AI - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

What is the main role of a workflow orchestrator in multi-agent AI systems?

Imagine you have several AI agents working together to complete a complex task. What does the workflow orchestrator do?

A It manages the order and timing of tasks among agents to ensure smooth collaboration.

B It trains each agent individually without coordinating their tasks.

C It only collects data from agents but does not control their actions.

D It replaces all agents by performing all tasks itself.

💡 Hint

Think about how a conductor leads musicians in an orchestra.

The workflow orchestrator acts like a conductor, coordinating when and how each agent performs its part to complete the overall task efficiently.

❓ Model Choice

intermediate

Which model architecture best supports workflow orchestration across multiple AI agents?

You want to design a system where multiple AI agents communicate and coordinate their actions. Which architecture is most suitable?

A Centralized controller that sends commands and receives feedback from agents.

B Independent agents working in isolation without communication.

C Single monolithic model handling all tasks without delegation.

D Random task assignment without any coordination.

💡 Hint

Think about a manager who assigns tasks and checks progress.

A centralized controller can orchestrate tasks by sending commands and receiving feedback, enabling coordination among agents.

❓ Metrics

advanced

Which metric best measures the efficiency of workflow orchestration across agents?

You want to evaluate how well your orchestrator coordinates multiple agents to complete a task quickly and correctly. Which metric should you use?

A Memory usage of each agent individually.

B Number of agents in the system.

C Total task completion time combined with task success rate.

D Number of lines of code in the orchestrator.

💡 Hint

Think about measuring speed and correctness together.

Combining total completion time and success rate shows how efficiently and accurately the orchestrator manages the workflow.

🔧 Debug

advanced

Why does this orchestrator code cause deadlock when coordinating agents?

Consider this simplified pseudocode for an orchestrator:

while not all_tasks_done:
    for agent in agents:
        if agent.waiting_for_response:
            continue
        task = get_next_task(agent)
        send_task(agent, task)
        wait_for_agent_response(agent)

Why might this cause a deadlock?

A Because agents never receive any tasks.

B Because the orchestrator waits for each agent's response before sending tasks to others, blocking progress.

C Because the loop never checks if all tasks are done.

D Because the orchestrator sends tasks too quickly without waiting.

💡 Hint

Think about waiting for one agent blocking others.

The orchestrator waits for each agent's response before moving on, so if an agent is stuck, all others wait indefinitely, causing deadlock.

❓ Predict Output

expert

What is the output of this multi-agent orchestration simulation?

Given the following Python code simulating two agents coordinated by an orchestrator, what is printed?

Agentic AI

class Agent:
    def __init__(self, name):
        self.name = name
        self.tasks_done = 0
    def perform_task(self):
        self.tasks_done += 1
        return f"{self.name} done task {self.tasks_done}"

class Orchestrator:
    def __init__(self, agents):
        self.agents = agents
    def run(self):
        results = []
        for _ in range(2):
            for agent in self.agents:
                results.append(agent.perform_task())
        return results

agents = [Agent("A1"), Agent("A2")]
orch = Orchestrator(agents)
output = orch.run()
print(output)

A ["A2 done task 1", "A1 done task 1", "A2 done task 2", "A1 done task 2"]

B ["A1 done task 1", "A1 done task 2", "A2 done task 1", "A2 done task 2"]

C ["A1 done task 1", "A2 done task 1"]

D ["A1 done task 1", "A2 done task 1", "A1 done task 2", "A2 done task 2"]

💡 Hint

Look at the nested loops: outer loop runs twice, inner loop runs over agents.

The outer loop runs 2 times, and for each iteration, both agents perform a task in order, so the output lists tasks done in sequence.