NCP-AAI practice question 3 of 25

Which two orchestration methods are MOST suitable for implementing complex agentic workflows that require both external data access and specialized task delegation? (Choose two.)

Full set: 118 questions
Answer options
A: Agentic orchestration with specialized expert system delegation Correct answer
B: Prompt chaining to accomplish state management
C: Manual workflow coordination without automation
D: Retrieval-based orchestration for external data Correct answer
E: Static rule-based routing with predefined pathways
Correct Answer:
A. Agentic orchestration with specialized expert system delegation; D. Retrieval-based orchestration for external data
Explanation

Implementing complex agentic workflows requires both structured task management and access to external knowledge. Agentic orchestration with specialized expert system delegation (A) provides a framework where a primary agent can route specific, complex tasks to other agents with specialized skills, fulfilling the delegation requirement. Retrieval-based orchestration (D), often implemented via Retrieval-Augmented Generation (RAG), directly addresses the need for external data access by enabling the system to fetch and incorporate information from external knowledge bases, ensuring the workflow's outputs are grounded in factual, up-to-date data.

Why Incorrect

B. Prompt chaining is a technique for structuring a single model's reasoning process, not a comprehensive method for orchestrating multiple agents or external data.
C. Manual workflow coordination is the antithesis of an automated agentic system and is not a scalable or efficient solution for complex tasks.
E. Static rule-based routing is too rigid for complex, dynamic workflows that often require adaptive decision-making based on context.

References

1. Lewis, P., et al. (2020). Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. Advances in Neural Information Processing Systems, 33, 9459-9474. (This paper introduces the RAG model, the foundational concept for retrieval-based orchestration).

2. Wu, T., et al. (2023). AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation. arXiv preprint arXiv:2308.08155. (Section 2.1 discusses how orchestrators can manage conversations between specialized agents to solve complex tasks).

3. Stanford University. (2023). CS224N: NLP with Deep Learning, Lecture on Retrieval-Augmented LMs. (Course materials often cover RAG as a key architecture for grounding LLMs in external data).

Scroll to Top