AI series · Part 4: Agentic Automation
AI Agents: From Chatbots That Talk to Assistants That Act
A chatbot tells you what to do. An agent plans, calls tools, checks results and repeats until the job is done. Here's a real agent trace and the guardrails that keep it safe.
A chatbot only talks. An agent can plan and take actions in a loop: think → use a tool → check the result → repeat, until the job is done.
A chatbot is like customer care that only tells you what to do. An agent is like a personal assistant who does it for you.
A real-world example
An IT helpdesk agent gets "I forgot my password". It checks who the user is, resets the password, updates the ticket and emails the user, with no human needed.
An agent trace from our HR assistant
Task: "Rahul wants 2 days of sick leave."
1. search_policies("sick leave") → 12 days/year, after probation
2. lookup_employee("Rahul") → balance 8, manager Priya
3. apply_leave(Rahul, 2 days) → sent to Priya for approvalIf Rahul asks for 20 days, the tool itself refuses: "Rejected: only 8 days left." The rule is enforced in code, not left to the model's judgement.
Guardrails
- A step limit, so a confused agent can't loop forever.
- Checks inside the tools, like the leave balance check above.
- Human approval for risky actions. Here, the manager approves the leave.
An agent is an LLM in a loop that chooses and calls tools to finish a task. I add guardrails: a step limit, checks inside the tools, and human approval for risky actions.
- The short version
Next in the GenAI series: Tool Calling: How LLMs Use Your APIs Without Running Code
Start from the top: GenAI Quick Answers and Series Guide
We deploy these patterns across production engagements in fintech, healthcare, and eCommerce. Talk to a senior engineer about how they apply to your platform.