<aside> 🦥

Sources

https://www.anthropic.com/engineering/building-effective-agents

https://cdn.openai.com/business-guides-and-resources/a-practical-guide-to-building-agents.pdf

</aside>

<aside> 🦥

Related notes: for modern multi-agent setups see Subagents & Multi-Agent Orchestration, and for single-agent reasoning shapes see AI Agent Architectures.

</aside>

LLM Agent Workflows (From OpenAI)

Basic Types

• Workflows: LLMs following fixed steps
• Agents: LLMs choosing their own path and tools

Main Ideas

• Start simple
• Add complexity only when needed
• Think about speed vs. cost vs. quality
• Know how things work under the hood
• Keep everything clear and fixable

Five Workflow Types

1. Prompt Chaining

• Break tasks into steps that happen one after another
• Check results between steps
• Good for: Tasks with clear steps
• Examples: Writing content then translating it

2. Routing

• Sort input and send to the right process