All articles
Engineering guides for LangChain chains and agents: RAG pipelines, retrievers, tool calling, structured output and token cost control.
-
How to Stream LangChain Responses: Models, Agents, APIs
Stream LangChain output with model.stream(), agent stream_mode messages, and astream_events, then ship tokens over SSE without proxy buffering.
-
LangChain Chunk Size: Configure Splitters for RAG
Configure LangChain chunk_size and chunk_overlap, choose character or token units, preserve source metadata, and compare splitter settings for RAG.
-
LangChain Memory Types: Legacy Classes and Replacements
This guide compares legacy LangChain memory classes with 1.x checkpointers, stores, trimming, and summarization for short- and long-term memory.
-
LangChain with Ollama: Local Setup, Tools, and RAG
This guide explains how LangChain connects to Ollama for local chat, streaming, tool calling, structured output, embeddings, and RAG.
-
LangChain Agent Errors: Loops, Tools, Parsing
Why LangChain agents loop forever, skip tools, pass bad arguments or fail to parse output, and the configuration changes that fix each symptom.
-
LangChain vs LangGraph: Agents, State and Control Flow
Compare LangChain vs LangGraph for agent loops, state, checkpoints and human approval, with guidance on when to use create_agent or a custom graph.
-
LangChain RAG Pipeline: Setup to First Answer
How a LangChain RAG pipeline fits together: loading, chunking, embeddings, vector storage and retrieval, plus the settings that decide answer quality.
-
LangChain Building Blocks: Chains, Tools and Agent Control
What LangChain actually abstracts, when an agent loop is the wrong choice, and how to keep retrieval, memory and token cost under control in production.