Adaptive Retrieval With Query Routing

via “multi-index query orchestration with hybrid retrieval strategies”

LlamaIndex is the leading document agent and OCR platform

Unique: Implements composable QueryEngine routers that can combine vector, keyword, graph, and structured queries through a unified interface with pluggable result merging strategies. Unlike LangChain's retriever composition (which chains retrievers sequentially), LlamaIndex's QueryEngine supports parallel multi-index querying with configurable fusion algorithms.

vs others: Enables true hybrid search with automatic result normalization and ranking, whereas LangChain requires manual result merging and score normalization across different retriever types.

via “adaptive-retrieval-with-query-routing”

This repository showcases various advanced techniques for Retrieval-Augmented Generation (RAG) systems. Each technique has a detailed notebook tutorial.

Unique: Implements query-aware routing that dynamically selects retrieval strategies based on query characteristics, allowing different query types to use optimized methods rather than forcing all queries through a single pipeline — an adaptive approach that improves both efficiency and quality

vs others: More efficient than applying all retrieval strategies to every query (fusion) because it selects the most appropriate strategy, and more effective than single-strategy systems because it adapts to query type

via “semantic document retrieval with query routing”

AI PDF chatbot agent built with LangChain & LangGraph

Unique: Implements explicit query routing as a LangGraph node rather than always retrieving — this reduces unnecessary vector DB queries and latency for general-knowledge questions. Routes via LLM decision logic (not keyword heuristics), enabling nuanced routing for complex queries.

vs others: More efficient than always-retrieve RAG patterns because it skips vector search for non-document queries; more flexible than rule-based routing because LLM routing adapts to query semantics rather than fixed keywords.

via “query controller with retrieval and llm integration”

RAG (Retrieval Augmented Generation) Framework for building modular, open source applications for production by TrueFoundry

Unique: Implements pluggable Query Controllers that orchestrate the full RAG pipeline (embedding generation → vector search → optional reranking → LLM inference) with support for different retrieval strategies and streaming responses. Integrates with Model Gateway for both embedding and LLM access, allowing strategy and model changes through configuration.

vs others: More modular than monolithic RAG chains (allowing strategy swapping) and more transparent than black-box RAG APIs (showing retrieval results and reasoning), enabling teams to debug and optimize each pipeline stage independently.

via “rag-sql hybrid query routing with semantic-to-sql translation”

In-depth tutorials on LLMs, RAGs and real-world AI agent applications.

Unique: Implements intelligent semantic-to-SQL routing using Cleanlab Codex rather than rule-based heuristics, enabling context-aware decisions about which retrieval path to use based on query intent and available data sources

vs others: More accurate than regex/keyword-based routing and faster than naive dual-retrieval approaches because it makes a single intelligent routing decision upfront rather than executing both paths and merging results

via “adaptive rag with query routing and dynamic context selection”

Ready-to-run cloud templates for RAG, AI pipelines, and enterprise search with live data. 🐳Docker-friendly.⚡Always in sync with Sharepoint, Google Drive, S3, Kafka, PostgreSQL, real-time data APIs, and more.

Unique: Implements query routing as a first-class pipeline component that dynamically selects retrieval strategies based on query classification, enabling cost and latency optimization without sacrificing answer quality. Supports both rule-based routing (fast, deterministic) and LLM-based routing (flexible, learned).

vs others: More sophisticated than basic RAG for high-volume systems; avoids the overhead of always retrieving context. Pathway's dataflow engine enables efficient routing without external orchestration frameworks.

via “semantic-routing-with-learned-gnn-optimization”

AgentDB v3 - Intelligent agentic vector database with RVF native format, RuVector-powered graph DB, Cypher queries, ACID persistence. 150x faster than SQLite with self-learning GNN, 6 cognitive memory patterns, semantic routing, COW branching, sparse/part

Unique: GNN-based routing learns from agent's own query patterns rather than using static heuristics — routing weights adapt to domain-specific characteristics and evolve as agent's knowledge base grows

vs others: More adaptive than fixed routing rules, and more efficient than querying all memory patterns in parallel — learns which patterns are most useful for specific query types

via “routing pattern for dynamic task direction based on query classification”

Agentic-RAG explores advanced Retrieval-Augmented Generation systems enhanced with AI LLM agents.

Unique: Implements routing as an intelligent classification step that analyzes query characteristics to select specialized handlers, rather than using static rules or random assignment, enabling adaptive pipeline selection based on query semantics.

vs others: More efficient than single-pipeline systems by avoiding unnecessary processing steps, and more adaptive than rule-based routing by using LLM reasoning to classify queries based on semantic content.

via “query engine with multi-stage retrieval and reranking”

Interface between LLMs and your data

Unique: Implements multi-stage retrieval pipeline with pluggable rerankers and response synthesis modes, supporting query decomposition (SubQuestionQueryEngine) and routing (RouterQueryEngine) without requiring custom orchestration code. Integrates reranking as a first-class abstraction rather than post-processing.

vs others: More sophisticated than basic vector search by supporting reranking, query decomposition, and response synthesis in a unified pipeline; enables complex multi-hop queries and improves answer quality through multi-stage filtering.

via “streamlined retrieval of findings”

Search leaked databases for email addresses, phone numbers, usernames, domains, and other identifiers. View categorized results across multiple sources to pinpoint relevant exposures. Speed investigations with targeted lookups and streamlined retrieval of findings.

Unique: Incorporates a context-aware suggestion engine that enhances retrieval speed by leveraging recent search history.

vs others: Faster retrieval than standard search tools, which require full re-querying of databases.

via “customizable query routing”

MCP server: db-map

Unique: Features a dynamic routing engine that evaluates query characteristics in real-time, allowing for optimized database interactions.

vs others: More flexible than static routing mechanisms, as it adapts to the nature of each query rather than applying a one-size-fits-all approach.

via “adaptive rag with query-dependent retrieval strategy selection”

Open-source Python library to build real-time LLM-enabled data pipeline.

Unique: Dynamically selects retrieval strategy based on query analysis, eliminating need for manual strategy selection. Integrates query analysis into the retrieval pipeline, enabling intelligent routing without separate preprocessing steps.

vs others: More effective than fixed retrieval strategies because it adapts to query characteristics; more efficient than trying all strategies because it selects the best one upfront.

via “search-intent-recognition-and-routing”

GPT-4o mini Search Preview is a specialized model for web search in Chat Completions. It is trained to understand and execute web search queries.

Unique: Search routing is embedded as a learned behavior in the model's forward pass rather than implemented as a separate classifier or rule engine, allowing the model to make context-aware routing decisions that account for conversation history and nuanced query phrasing

vs others: More efficient than always-on search (vs. Perplexity or traditional RAG systems) because the model learns to skip unnecessary searches, reducing latency and API costs while maintaining factual accuracy on time-sensitive queries