What are AI Agents, Really? › Anatomy of an Agent

Let's examine the components. ## Components | Component | Role | |------------------|----------------------------------------------------------------------| | **Input** | A user message or external event that starts the agent | | **LLM Call** | Interprets input, tracks memory, suggests next steps | | **Tool Call** | Carries out real-world actions beyond what an LLM can do | | **Control Logic**| Drives state transitions through branching, loops, and end checks | | **Memory/State** | Holds context, decisions, and history | ## Simplified Execution Flow > **User Prompt:** "Plan me a vacation to Italy." 2. **LLM State 1:** Ask for travel dates and budget 3. **Tool State 2:** Look up hotels based on input 4. **LLM State 3:** Recommend options 5. **LLM State 4:** Ask for confirmation 6. **Tool State 5:** Book the hotel and complete Each numbered step is a state implemented as a function. Transitions depend on LLM and tool outputs.