Preview — full styling will appear after the next deploy completes.

agentic-ai-patterns

Prompt Chaining

Linear sequential pipeline

The simplest agentic pattern: the output of one LLM call becomes the input of the next. No loops, no branching — a clean assembly line for structured text transformations.

flowchart LR
    S([__start__]) --> X[extract]
    X --> A[analyse]
    A --> F[format_report]
    F --> Z[summarise]
    Z --> E([__end__])

Prompt Chaining is the foundational building block for structured workflows. Each step in the chain has a single, clearly defined responsibility: extract relevant data, analyse it, format a structured report, then summarise it for different audiences. The output of each step flows directly into the next as context.

Unlike ReAct or reflection loops, there are no tools, no conditional branches, and no iteration — just a linear pipeline of LLM calls. This makes it the easiest pattern to reason about, debug, and monitor. Every intermediate state is inspectable in the graph's state object.

Use Prompt Chaining when you have a well-understood transformation pipeline where each stage's requirements are stable. Typical applications include document processing (extract → summarise → classify), content generation (outline → draft → edit), and data enrichment (parse → enrich → format).