LangChain is a Python framework for applications that connect large language models with prompts, data sources, tools, and workflow logic. It simplifies the LLM application lifecycle and provides a common interface across providers, including a shared invoke() method for sending prompts. section 9.3.1
LangChain supports two useful levels of integration:
- One-shot prompting wraps provider access in modules such as ChatOpenAI, while retaining configurable models, temperature, tokens, retries, and timeouts. section 9.3.1
- Agents connect tools and data sources to an LLM, enabling workflows such as retrieving current financial news before answering a question. section 9.3.1
