# RAG & Memory

> RAG pipelines and agent memory: frameworks like LangChain, LlamaIndex, and DSPy, rerankers like Cohere Rerank and ColBERT, document parsers like Unstructured and Docling, and memory layers like Mem0 and Letta that give agents persistence across sessions. Retrieval quality is won in parsing and reranking, so start there before switching frameworks.

21 tools in 4 subcategories. Web page: https://ailandscape.org/category/rag-memory · Each tool also has a markdown version at /tool/{slug}.md

## LLM Frameworks & RAG

- [DSPy](https://ailandscape.org/tool/dspy): Framework for algorithmically optimizing LM prompts and pipelines
- [Haystack](https://ailandscape.org/tool/haystack): End-to-end NLP framework for LLM apps
- [LangChain](https://ailandscape.org/tool/langchain): Framework for developing LLM-powered applications
- [LlamaIndex](https://ailandscape.org/tool/llamaindex): Data framework for LLM applications
- [Semantic Kernel](https://ailandscape.org/tool/semantic-kernel): Microsoft's SDK for AI orchestration
- [Spring AI](https://ailandscape.org/tool/spring-ai): Spring framework for Java AI application development

## Retrieval & Reranking

- [BGE Reranker](https://ailandscape.org/tool/bge-reranker): BAAI's high-performance multilingual reranker
- [Cohere Rerank 3](https://ailandscape.org/tool/cohere-rerank-3): State-of-the-art reranking model for RAG relevance
- [ColBERT](https://ailandscape.org/tool/colbert): Late interaction retrieval model with token-level embeddings
- [FlashRank](https://ailandscape.org/tool/flashrank): Ultra-lite cross-encoder reranking for RAG
- [Jina Reranker](https://ailandscape.org/tool/jina-reranker): Multilingual reranker for improving RAG precision

## Document Parsing

- [Docling](https://ailandscape.org/tool/docling): IBM's fast document conversion and parsing toolkit
- [LlamaParse](https://ailandscape.org/tool/llamaparse): Optimized document parsing for RAG by LlamaIndex
- [Marker](https://ailandscape.org/tool/marker): Convert PDFs to markdown quickly and accurately
- [Reducto](https://ailandscape.org/tool/reducto): High-accuracy document extraction for complex PDFs
- [Unstructured](https://ailandscape.org/tool/unstructured): Document processing for RAG pipelines

## Memory Layer

- [Cognee](https://ailandscape.org/tool/cognee): Semantic memory layer building knowledge graphs for AI agents
- [Mem0](https://ailandscape.org/tool/mem0): Universal persistent memory layer for AI agents and assistants
- [Letta (MemGPT)](https://ailandscape.org/tool/letta-memgpt): OS-inspired memory management for LLMs enabling unbounded virtual context
- [supermemory](https://ailandscape.org/tool/supermemory): Memory and context API that gives AI agents long-term recall
- [Zep](https://ailandscape.org/tool/zep): Temporal knowledge graph memory for agents (Graphiti)

---

Part of [AI Landscape](https://ailandscape.org), an open map of the AI ecosystem. Index for AI assistants: https://ailandscape.org/llms.txt