AI agent
AI agent: An AI agent is a program that uses a language model to understand a goal, decide which steps to take and carry them out with external tools, such as a calendar, a CRM or email, until it finishes the task or hands it to a person.
An AI agent is a program that receives a goal, understands it with a language model and completes it using tools. If a customer writes “I need to move my Thursday appointment”, the agent finds the appointment in your calendar, checks the free slots, offers two options and confirms the one the customer picks. Nobody on your team has to step in.
How does an AI agent work?
An agent combines three pieces:
- A language model (LLM), such as GPT, Claude or Gemini. It understands the request and decides the next step.
- Tools: functions the agent can call. Check a calendar, read an order, create a CRM contact or send an email.
- Instructions and context: what it may do, what it may not, what tone it uses and when it must pass the conversation to a person.
The agent works in a loop. It reads the request, picks a tool, looks at the result and decides again. It repeats until the task is done or until it detects that it needs human help.
To answer with your company’s data and not with general knowledge, agents usually rely on RAG: before replying, they search your documents, price lists or policies and answer only with what they find.
What is the difference between an AI agent and a chatbot?
A classic chatbot follows a tree of options. If the customer asks something unplanned, it gets stuck or returns a generic message. Chatbots built on generative AI understand questions better, but they still only reply.
An agent acts. You see the difference in the outcome: the chatbot says “you can book on our website” and the agent books it for you.
| Chatbot | AI agent | |
|---|---|---|
| Understands free-form questions | Depends | Yes |
| Checks your systems | No | Yes |
| Runs actions | No | Yes |
| Hands off to a person | Sometimes | Yes, with the conversation context |
What does a business use an AI agent for?
Agents pay off most in repetitive, high-volume tasks with clear rules. The most common uses in small and mid-sized businesses:
- Call answering: pick up the phone 24/7, resolve frequent questions and book appointments. This is the AI receptionist use case.
- WhatsApp support: reply to customers, send quotes and follow up.
- Sales qualification: ask each lead the right questions, score it and book a call with sales.
- Internal assistant: answer your team with company procedures, price lists or contracts.
- Admin tasks: read invoices, extract the data and record it in the accounting software.
Example: an agent at a dental practice
Picture a dental practice with two chairs and a receptionist who cannot keep up. On weekdays about 40 calls come in per day, and when she is with a patient, many go unanswered.
The practice puts a voice agent on its usual phone number. The agent picks up, works out whether the caller is already a patient, checks the calendar in the practice management software and offers slots. If someone calls with severe pain, the agent flags it as urgent and messages the receptionist. Price questions get answered from the fee list the practice provided, and if someone asks for a person, the agent transfers the call.
The receptionist stops answering the phone to reschedule appointments and spends that time on the patients in the waiting room.
What do you need to have an AI agent?
- A specific process: “answer WhatsApp outside office hours” works; “improve customer service” is too broad.
- The information it will draw on: FAQs, price lists, policies and catalog. It does not need to be tidy.
- Access to the tools: calendar, CRM, ERP or online store, with permissions limited to what is needed.
- A person to review its answers during the first weeks.
When an agent coordinates several steps and decides the order itself, people talk about agentic AI. To see how we build agents and what they cost in your case, read AI agent development for the full process, or tell us about your case.