LangChain vs Rasa

LangChain and Rasa, despite sounding like a pair of intergalactic bounty hunters, are actually both open-source frameworks designed to help humans communicate with machines in a slightly less frustrating manner. They each dabble in large language models, allow for customization and come with enough documentation to make a Vogon poet weep with joy. Both can be used to create chatbots, virtual assistants and other AI-powered contraptions that, with a bit of fine-tuning, might just understand what you’re asking them to do. This, of course, is a vast improvement over most existing automated phone menus.

LangChain, born in the United States in 2022, is the younger of the two and is quite obsessed with chaining together large language models in creative and occasionally terrifying ways. It is designed for AI developers who enjoy making complex, reasoning-based systems that can pull in data from all sorts of places and then, ideally, produce something useful. Its real talent is orchestrating structured and unstructured data retrieval, which sounds very technical but mostly means it can dig through heaps of information without having a nervous breakdown.

Rasa, on the other hand, hails from Germany and has been brooding over the nature of conversation since 2017. Unlike LangChain’s "let’s throw LLMs at everything" approach, Rasa is more concerned with building rule-based and machine-learning-driven chatbots that don’t require an internet connection to function. It specializes in understanding user intent, recognizing entities and managing multi-turn dialogues—all crucial skills if you ever need a chatbot to talk to you about your insurance policy without making you want to hurl your phone into a black hole.

See also: Top 10 Chatbot Builders