---
title: "Building Effective AI Agents"
description: "Effective agentic systems need not begin with complex frameworks: the post reports that successful implementations commonly use simple, composable patterns and recommends choosing the least complex solution that meets the task's needs. It distinguishes workflows, which follow predefined code paths, from agents, where LLMs dynamically direct their processes and tool use. The proposed foundation is an augmented LLM with retrieval, tools, and memory, potentially integrated through the Model Context Protocol; larger designs include prompt chaining, routing, parallelization, and orchestrator-workers patterns. Frameworks can simplify implementation but may obscure prompts and responses, so direct API use and understanding underlying code are recommended. Tool definitions are treated as an interface requiring testing, clear examples, low formatting overhead, and mistake-resistant parameters; absolute filepaths eliminated observed errors in a coding agent."
---

# Building Effective AI Agents

[Anthropic](https://yomu.fyi/company/anthropic) · Dec 19, 2024

**Type:** Explainer

## Summary

Effective agentic systems need not begin with complex frameworks: the post reports that successful implementations commonly use simple, composable patterns and recommends choosing the least complex solution that meets the task's needs. It distinguishes workflows, which follow predefined code paths, from agents, where LLMs dynamically direct their processes and tool use. The proposed foundation is an augmented LLM with retrieval, tools, and memory, potentially integrated through the Model Context Protocol; larger designs include prompt chaining, routing, parallelization, and orchestrator-workers patterns. Frameworks can simplify implementation but may obscure prompts and responses, so direct API use and understanding underlying code are recommended. Tool definitions are treated as an interface requiring testing, clear examples, low formatting overhead, and mistake-resistant parameters; absolute filepaths eliminated observed errors in a coding agent.

## Context

The post draws on work with dozens of teams building LLM agents across industries and addresses how to choose between simpler LLM applications, workflows, and dynamically directed agents. It notes that agentic systems can trade latency and cost for better task performance, while frameworks may obscure underlying prompts and responses.

## Approach / What changed

It presents an augmented LLM as the foundation, then describes composable workflow patterns including prompt chaining, routing, parallelization, and orchestrator-workers. It recommends starting with direct LLM APIs, using frameworks only with an understanding of their underlying code, and designing, testing, and simplifying tool interfaces.

## Takeaways

- Workflows use predefined orchestration, while agents let LLMs dynamically control their processes and tool usage.
- Tool specifications need prompt-engineering attention: clear descriptions, examples, edge cases, low formatting overhead, and parameters designed to prevent mistakes.
- In a coding-agent implementation, requiring absolute filepaths instead of relative ones fixed filepath errors after the agent moved away from the root directory.

**Tags:** [AI Agents](https://yomu.fyi/topic/ai-agents), [MCP](https://yomu.fyi/topic/mcp), [Prompt Engineering](https://yomu.fyi/topic/prompt-engineering)

- Source: [Anthropic](https://www.anthropic.com/engineering/building-effective-agents)
- Source URL: https://www.anthropic.com/engineering/building-effective-agents
- Ingested by Yomu: 2026-09-09T11:36:40.950Z

[Read original post](https://www.anthropic.com/engineering/building-effective-agents)