# Agentic Workflow An **agentic workflow** is a structured process in which an AI [agent](agent) autonomously plans and executes a series of steps to complete a complex task. Rather than relying on a human to direct every action, the agent decides its own path through the work — using tools, evaluating results, and adapting as needed. ## Components of an Agentic Workflow A typical agentic workflow consists of: - **Task definition** — A goal or instruction provided to the agent (e.g., "add authentication to this API") - **Planning** — The agent breaks the goal into sub-tasks - **Tool execution** — The agent calls tools such as code editors, shells, APIs, or search - **Observation** — The agent reviews results and adjusts its plan - **Completion or escalation** — The agent signals done, or asks a human for guidance ## Why Agentic Workflows Need Human Oversight Agents can make mistakes — misunderstand requirements, take irreversible actions, or get stuck in loops. **[Human-in-the-loop](human-in-the-loop)** checkpoints prevent small errors from cascading into large problems. With [AgentRQ](/), you can define exactly when the agent should pause and wait for your approval before proceeding. ## Single-Agent vs. Multi-Agent Workflows | Pattern | Description | |---------|-------------| | Single agent | One agent handles the entire workflow end-to-end | | Multi-agent | Multiple specialized agents collaborate, passing results to each other | | Supervisor pattern | An orchestrator agent delegates tasks to sub-agents | ## Agentic Workflows with Claude Code [Claude Code](claude-code) is purpose-built for agentic coding workflows. It can autonomously read codebases, write tests, fix bugs, and open pull requests — while integrating with AgentRQ to notify you at critical decision points via [MCP](mcp). ## Related Terms - [Agent](agent) - [Autonomous Agent](autonomous-agent) - [Human-in-the-Loop](human-in-the-loop) - [Task](task) - [Workflow](workflow)