liveBook · Manning

Sequential agent structure

As explained in Grokking AI Applications

An agentic structure in which multiple agents execute in a fixed order, passing each agent's output to the next agent.

A sequential agent structure connects multiple specialized agents in a fixed order. One agent receives the request, uses its tools, and passes its output to the next agent, creating a predictable pipeline. section 6.2

Use the pattern when each stage naturally depends on the previous stage:

  • Assign each agent a focused role and its own tools. section 6.2
  • Pass the first agent’s result directly to the next agent. section 6.2
  • Choose the pattern for workflows requiring a fixed order, such as drafting a document and then formatting or reviewing it. section 6.2
  • Accept that later agents must wait, giving up the flexibility of conditional delegation. section 6.2

HealthyChef™

A chef agent searches a 1919 Italian cookbook for a recipe matching the user’s request or available ingredients. It passes the result to a dietitian agent, which evaluates the recipe against NHS EatWell guidelines. Retrieval-augmented generation and a vector database support the chef’s search, while the dietitian uses a guidelines-access tool. section 6.3

Figure 6.2 The four types of agentic structures: How would you organize your agents?

Compare sequential, single, hierarchical, and other agentic structures through an organizational overview. section 6.2

How Grokking AI Applications covers it

How deep it goes
Explains the fixed handoff, control benefit, waiting cost, and application design through a chef-and-dietitian workflow.
Examples
Uses document drafting and review, plus HealthyChef’s cookbook search followed by nutritional evaluation.
What it assumes
Assumes familiarity with agents and tools while demonstrating retrieval-augmented generation and vector search within the workflow.
What you can do after
Build it from scratch by assigning agent roles, connecting tools, and passing outputs through a predetermined chain.
How it connects
Connects sequential agents with single agents, hierarchical agents, specialized tools, retrieval-augmented generation, and vector databases.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Grokking AI Applications. The section links point to the parts of the book it draws on.