Loading patterns…

⚡

Advanced RAG(ARAG)

Enhanced RAG with pre-retrieval and post-retrieval optimizations including query expansion, reranking, and context curation

Complexity: mediumKnowledge Retrieval (RAG)

In 30 seconds

What: Retrieves documents using query expansion, hybrid sparse-dense search, reranking with cross-encoders, and curated context assembly before feeding to the model.
When to use: Knowledge-heavy tasks where retrieval quality directly impacts answer accuracy, especially with large or noisy document collections.
Watch out: Compounding latency from multiple retrieval passes and reranking stages can make response time unacceptable for real-time applications.

Ask the AI expert about this pattern

Opens the assistant with your question prefilled. You review it before sending.

Get this catalog as an MCP server Tune a prompt for this pattern

Enhanced RAG with pre-retrieval and post-retrieval optimizations including query expansion, reranking, and context curation

Pre-retrieval optimization
Query expansion techniques
Post-retrieval processing
Context curation
Document reranking
Quality assessment
Hybrid sparse-dense fusion with reciprocal rank fusion (RRF)
Late-interaction retrieval (ColBERT-style) and cross-encoder reranking

The Agent Architect

One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.

Weekly email, one-click unsubscribe. We only use your address to send the briefing.

References

The papers, specifications, and repositories this pattern is based on.

Retrieval-Augmented Generation for Large Language Models: A Survey (Gao et al., 2023)arXiv:2312.10997
Seven Failure Points When Engineering a Retrieval Augmented Generation System (Barnett et al., 2024)arXiv:2401.05856
RAG vs Fine-tuning: Pipelines, Tradeoffs, and a Case Study (Ovadia et al., 2023)arXiv:2401.08406
A Comprehensive Survey of RAG: Evolution and Future Directions (Gupta et al., 2024)arXiv:2410.12837
Precise Zero-Shot Dense Retrieval without Relevance LabelsarXiv:2212.10496

From the engineer behind this catalog

Get your agent architecture reviewed

This page documents one pattern. Your system runs dozens, and most failures live in how they fit together. Have the whole design reviewed against the 288 patterns in this catalog: architecture, reliability, evaluation and cost, every finding mapped to the pattern that fixes it.

€750 instead of €1,500, one week, written report and walkthrough call, until 30 September

See the review

Patterns

Loading patterns…

Design Patterns & Techniques

Loading patterns…

⚡

Advanced RAG(ARAG)

Enhanced RAG with pre-retrieval and post-retrieval optimizations including query expansion, reranking, and context curation

Complexity: mediumKnowledge Retrieval (RAG)

In 30 seconds

What: Retrieves documents using query expansion, hybrid sparse-dense search, reranking with cross-encoders, and curated context assembly before feeding to the model.
When to use: Knowledge-heavy tasks where retrieval quality directly impacts answer accuracy, especially with large or noisy document collections.
Watch out: Compounding latency from multiple retrieval passes and reranking stages can make response time unacceptable for real-time applications.

Ask the AI expert about this pattern

Opens the assistant with your question prefilled. You review it before sending.

Get this catalog as an MCP server Tune a prompt for this pattern

30-Second Overview

Pattern: Enhanced retrieval pipeline with query preprocessing, multi-stage retrieval, neural reranking, and context optimization

Why: Addresses limitations of naive RAG through query understanding, relevance scoring, and context quality optimization

Key Insight: Pre-retrieval optimization + post-retrieval processing significantly improves accuracy and relevance

Quick Implementation

1Pre-process:Query expansion, rewriting, routing

2Multi-retrieve:Multiple retrieval strategies and sources

3Rerank:Neural rerankers (BGE, Cohere, etc.)

4Filter:Relevance scoring and context selection

5Generate:Context-optimized generation with citations

Example: expand_query → multi_retrieve → neural_rerank → filter_context → generate_with_citations

Do's & Don'ts

✅Implement query expansion (HyDE, query2doc)

✅Use neural rerankers (BGE-reranker, Cohere rerank)

✅Apply sentence window retrieval for context preservation

✅Implement relevance filtering with confidence thresholds

✅Use multiple embedding models for retrieval diversity

❌Skip query preprocessing and expansion

❌Rely solely on semantic similarity for ranking

❌Ignore document quality and freshness signals

❌Over-retrieve without proper filtering mechanisms

❌Neglect context window optimization

When to Use

Use When

• Production RAG systems requiring high accuracy
• Complex queries needing contextual understanding
• Large knowledge bases with noisy content
• Multi-domain or heterogeneous data sources
• Applications requiring source attribution

Avoid When

• Simple factual Q&A with clean data
• Resource-constrained environments
• Real-time applications (<100ms latency)
• Small knowledge bases with high-quality content
• Proof-of-concept or prototype systems

Key Metrics

Retrieval Precision

Relevant docs in top-k after reranking

Answer Faithfulness

Generated content grounded in retrieved docs

Context Relevance

Retrieved context relevance to query

Reranking Effectiveness

NDCG@k improvement vs base retrieval

Query Understanding

Semantic similarity after expansion/rewriting

End-to-End Latency

Including pre-processing and reranking overhead

Top Use Cases

Enterprise Search: Complex queries over large corporate knowledge bases with reranking

Legal Research: Multi-hop reasoning over case law with query expansion and relevance filtering

Medical Q&A: Clinical queries with domain-specific rerankers and confidence scoring

Technical Documentation: Developer queries with code-aware retrieval and context optimization

Research Assistant: Academic queries with citation tracking and multi-source retrieval

References & Further Reading

Deepen your understanding with these curated resources

Advanced RAG Surveys & Frameworks

Retrieval-Augmented Generation for Large Language Models: A Survey (Gao et al., 2023)

Seven Failure Points When Engineering a Retrieval Augmented Generation System (Barnett et al., 2024)

RAG vs Fine-tuning: Pipelines, Tradeoffs, and a Case Study (Ovadia et al., 2023)

A Comprehensive Survey of RAG: Evolution and Future Directions (Gupta et al., 2024)

Query Enhancement & Preprocessing

Precise Zero-Shot Dense Retrieval without Relevance Labels

Query2doc: Query Expansion with Large Language Models (Wang et al., 2023)

Query Rewriting for Retrieval-Augmented Large Language Models

Active Retrieval Augmented Generation (Jiang et al., 2023)

Neural Reranking & Relevance Scoring

M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillation

FlagEmbedding: Dense Retrieval and Retrieval-augmented LLMs (BAAI, 2023)

Cohere Rerank API Documentation and Best Practices

Context Optimization & Chunking

Lost in the Middle: How Language Models Use Long Contexts (Liu et al., 2023)

LongLLMLingua: Accelerating and Enhancing LLMs in Long Context Scenarios via Prompt Compression

Recursive Character Text Splitter: LangChain Documentation

Implementation Frameworks & Tools

LangChain Multi-Vector Retriever Implementation

Haystack Retriever Documentation

Weaviate Hybrid Search (Dense + Sparse) Documentation

Evaluation & Benchmarking

RAGAS: Automated Evaluation of RAG Applications

BEIR: A Heterogenous Benchmark for Zero-shot Evaluation of Information Retrieval Models

Benchmarking Large Language Models in Retrieval-Augmented Generation

Enhanced RAG with pre-retrieval and post-retrieval optimizations including query expansion, reranking, and context curation

Pre-retrieval optimization
Query expansion techniques
Post-retrieval processing
Context curation
Document reranking
Quality assessment
Hybrid sparse-dense fusion with reciprocal rank fusion (RRF)
Late-interaction retrieval (ColBERT-style) and cross-encoder reranking

The Agent Architect

One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.

Weekly email, one-click unsubscribe. We only use your address to send the briefing.

References

The papers, specifications, and repositories this pattern is based on.

Retrieval-Augmented Generation for Large Language Models: A Survey (Gao et al., 2023)arXiv:2312.10997
Seven Failure Points When Engineering a Retrieval Augmented Generation System (Barnett et al., 2024)arXiv:2401.05856
RAG vs Fine-tuning: Pipelines, Tradeoffs, and a Case Study (Ovadia et al., 2023)arXiv:2401.08406
A Comprehensive Survey of RAG: Evolution and Future Directions (Gupta et al., 2024)arXiv:2410.12837
Precise Zero-Shot Dense Retrieval without Relevance LabelsarXiv:2212.10496

From the engineer behind this catalog

Get your agent architecture reviewed

€750 instead of €1,500, one week, written report and walkthrough call, until 30 September

See the review

Loading patterns…

Advanced RAG(ARAG)

In 30 seconds

Ask the AI expert about this pattern

Advanced RAG: Overview

The Agent Architect

References

Get your agent architecture reviewed

Patterns

Loading patterns…

Design Patterns & Techniques

Loading patterns…

Advanced RAG(ARAG)

In 30 seconds

Ask the AI expert about this pattern

30-Second Overview

Quick Implementation

Do's & Don'ts

When to Use

Use When

Avoid When

Key Metrics

Top Use Cases

References & Further Reading

Advanced RAG Surveys & Frameworks

Query Enhancement & Preprocessing

Neural Reranking & Relevance Scoring

Context Optimization & Chunking

Implementation Frameworks & Tools

Evaluation & Benchmarking

Advanced RAG: Overview

The Agent Architect

References

Get your agent architecture reviewed

Patterns

Design Patterns & Techniques

Loading patterns…

Design Patterns & Techniques

30-Second Overview

Quick Implementation

Do's & Don'ts

When to Use

Use When

Avoid When

Key Metrics

Top Use Cases

References & Further Reading

Advanced RAG Surveys & Frameworks

Query Enhancement & Preprocessing

Neural Reranking & Relevance Scoring

Context Optimization & Chunking

Implementation Frameworks & Tools

Evaluation & Benchmarking

Design Patterns & Techniques