Prompt chaining is the process of using the output from one Large Language Model (LLM) call as input to another call in a sequence. It decomposes a task into smaller stages, allowing each prompt to focus on a narrower operation.
For example, one call may extract facts from a document, a second may organize them into an outline, and a third may draft a report. Different models can be used for different stages, and application code can validate intermediate results before continuing.
Prompt chaining is a common agentic workflow pattern, but a fixed chain is not necessarily an autonomous AI agent. The sequence is usually predefined rather than selected dynamically by the model.
Long chains increase latency and create opportunities for errors to propagate. Structured output, explicit validation, retries, and evaluation help make the boundaries between stages more reliable.
The LLM Knowledge Base is a collection of bite-sized explanations for commonly used terms and abbreviations related to Large Language Models and Generative AI.
It's an educational resource that helps you stay up-to-date with the latest developments in AI research and its applications.