TypeScript-Only AI Stack
“TypeScript version is a second-class citizen compared to the Python library”
Browser AI / Web Agent Stack
“RAG pipeline framework; the data source here is live web pages, not a document corpus”
LLM Production Infra Stack
“Full RAG framework; Qdrant's native clients handle retrieval without the extra abstraction”
Evaluation & Quality Stack
“Retrieval pipeline framework — evaluation here is model-and-prompt focused, not retrieval-focused”
Memory-Augmented Agent Stack
“RAG over documents; Mem0 handles episodic memory — what users said and decided, not what documents contain”
Graph RAG Stack
“LightRAG handles its own document ingestion and graph construction — adding LlamaIndex creates a redundant ingestion layer”