Agentic AI Reference Architecture

Knowledge & Data · Software component

Multi-Hop Answer Synthesizer

Software componentKnowledge & DataKnowledge & Dataarc:MultiHopAnswerSynthesizer

An answer synthesizer that combines sub-answers and supporting facts from multiple documents into a final answer attributed to its contributing sources.

Responsibility. Synthesizes sub-answers into a source-attributed final answer.

Also known as: Sub-answer synthesis, Research synthesis, Decomposed-context answer synthesizer, Multi-context synthesis

Variant of Answer Synthesizer abstract

When to choose. Choose when answers must combine sub-answers from multiple documents and cite the supporting sources for verification.

invokesspecializessends data tois orchestrated byinvokesis evaluated byis invoked byreceives data fromreceives data fromreceives data fromis configured byLLM Inference Service: invokesLLM Inference ServiceAnswer Synthesizer: specializesAnswer SynthesizerExplanation Presenter: sends data toExplanation PresenterRAG Query Orchestrator: is orchestrated byRAG Query OrchestratorConfidence Estimator: invokesConfidence EstimatorCitation Verifier: is evaluated byCitation VerifierMulti-Hop Retrieval Controller: is invoked byMulti-Hop Retrieval Cont…Cross-Source Consistency Verifier: receives data fromCross-Source Consistency…Parallel Sub-Query Retrieval Controller: receives data fromParallel Sub-Query Retri…Supporting Fact Extractor: receives data fromSupporting Fact ExtractorSynthesis Prompt Template: is configured bySynthesis Prompt Template
Direct neighbourhood (hover for relationship types)

Relationships

is configured by structural

invokes dependency

is invoked by dependency

receives data from dynamic

sends data to dynamic

is orchestrated by control

is evaluated by assurance

Design guidance

Quantitative guidance

As stated by the sources; verify before use.

Classification

Patterns
Source attributionHierarchical citation references (e.g., [1.1] = sub-query 1, chunk 1)Sectioned per-sub-query context
Quality attributes
Functional suitability: correctness and validity (ISO/IEC 25010 | NIST AI RMF: valid)Transparency and accountability (NIST AI RMF: accountable and transparent)
Risks mitigated
Hallucinated supporting evidenceFabricated comparisons or numbers

Sources

  1. Ch3.3: T. Nguyen, "Web Navigation and Interaction Benchmarks," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.3. ISBN: 9798244538229.
  2. Ch6.6: T. Nguyen, "Query Decomposition and Adaptive Retrieval," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 6.6. ISBN: 9798244538229.