Workflows vs Agents

Definition

Anthropic groups both under "agentic systems" but draws an architectural line: workflows orchestrate LLMs and tools through predefined code paths, whereas agents let LLMs dynamically direct their own processes and tool usage, maintaining control over how they accomplish tasks.

Key points

• Workflows — "systems where LLMs and tools are orchestrated through predefined code paths."
• Agents — "systems where LLMs dynamically direct their own processes and tool usage, maintaining control over how they accomplish tasks."
• Guiding principle: "find the simplest solution possible, and only increase complexity when needed."
• Agentic systems trade latency and cost for better task performance — don't reach for them by default.
• Many applications do NOT need agents; a single LLM call with retrieval and in-context examples is often enough.

Why it matters for the exam

• The workflow-vs-agent distinction is the framing that organizes every pattern below; expect scenarios asking you to pick a predefined path vs a self-directed agent.

Common gotchas

• Start simple; only add agentic complexity when it demonstrably improves outcomes.

See also

• Augmented LLM
• Autonomous Agents
• Prompt Chaining Workflow
• Agentic loop
• Task decomposition (chaining vs adaptive)

Sources

• blog-building-effective-agents