Updated October 2026 · 8 min read
Types of AI agents: watch all 5 answer the same customer message
Most articles list the agent types as definitions. That doesn't tell you which one to build. So here is one real message a clinic gets at 22:47, and how each type of agent handles it. The difference between type 1 and type 5 is the difference between a double booking and a client who pays you every month.
There are 5 classic types of AI agents: simple reflex (fixed rules), model-based reflex (rules + memory of the situation), goal-based (plans steps to reach a goal), utility-based (picks the best option) and learning agents (improve from feedback). Modern business agents are mostly goal-based agents built on a language model with tools, often working together in a multi-agent system.
Same request, 5 agent types
Tap a type to see its reply, or press Run all 5. It is a simulation: nothing is sent anywhere.
Pick an agent type above to see how it replies.
Same message, same model price. Only the agent design changed, and with it what a business will pay for.
See how a goal-based agent is built →The 5 types of AI agents, explained
The five types come from the classic AI textbook Artificial Intelligence: A Modern Approach (Russell & Norvig). Each type adds one ability to the one before it. IBM's overview uses the same five.
| Type | How it decides | Business example | Good for |
|---|---|---|---|
| 1. Simple reflex | If X, then Y. Only looks at the current message | Auto-reply with opening hours, keyword chatbot | Simple FAQs, very cheap |
| 2. Model-based reflex | Rules + an internal picture of the situation (memory, state) | Bot that knows your order or booking status | Status questions, support |
| 3. Goal-based | Plans the steps needed to reach a goal | Agent that books, moves or cancels appointments | Bookings, lead follow-up, admin |
| 4. Utility-based | Scores options and picks the best one | Picks the slot or offer with the best outcome for both sides | Scheduling, pricing, routing |
| 5. Learning | Improves its behaviour from feedback and outcomes | Agent that adds reminders after it sees no-shows | Anything that runs for months |
The modern types: what “AI agent” means in 2026
When people say “AI agent” today they rarely mean a textbook reflex agent. They mean one of these, built on a language model:
LLM agent with tools
- A language model that reads the request
- Calls tools: calendar, CRM, payments, email
- Plans the next step itself (goal-based)
- Can score options and learn from logs
- See 20 examples by industry
Multi-agent system
- Several specialised agents working together
- e.g. intake agent → booking agent → checker
- One agent per job keeps each one reliable
- Coordinated by an orchestrator
- See multi-agent systems explained
Other labels you will see are mostly about where an agent works, not how it decides: customer service agents, voice agents, coding agents, research agents. Under the hood, nearly all of them are goal-based LLM agents with tools, plus the infrastructure to run them reliably.
Which agent type does your client need?
Three questions about the business you want to build for.
Why the higher types are worth more to sell
A type 1 auto-reply is a one-off job. An agent that acts, picks the best option and gets better every month is a monthly service, because the business keeps needing it. Most AI agencies package it as a setup fee plus a monthly fee. Play with the numbers:
Example math before costs (model usage, tools, hosting, your time). An example, not a promise: your niche, price and delivery decide what you earn.
The free training shows how to build it and sell it as a monthly offer →
Next read: what are AI agents (start here) · how to build AI agents · how to start an AI automation agency · ways to make money with AI
Watch a goal-based agent being built — and sold monthly
Free training, in English, about 45–60 minutes: we build an agent that takes real actions for a business, then show how to package it and charge monthly for it. No coding background needed.
Watch the free training →Free · no credit card
Questions
What are the 5 types of AI agents?
Simple reflex, model-based reflex, goal-based, utility-based and learning agents. Each adds one ability: memory, planning, choosing the best option, and improving from feedback.
What is an intelligent agent in AI?
Anything that perceives its environment and acts on it to reach its goals. In business today that is usually a language-model agent that reads messages, uses tools and takes actions.
Which type of AI agent do businesses use most?
Goal-based agents built on a language model with tool access. Utility and learning behaviour is added when the business can measure what a good outcome is.
Is ChatGPT an AI agent?
A plain chat assistant mostly answers. It becomes an agent when it gets tools and a goal, so it can book, send or update things and decide its next step.
How do agencies make money with AI agents?
A one-time setup fee plus a monthly fee for hosting, monitoring and improving the agent.