Last updated: 2 November 2026
An AI agent is an AI tool that carries out a multi-step task on your behalf, using other tools such as your email, a spreadsheet or a CRM, and checks with you when it needs a decision. Generative AI creates content from a prompt. Agentic AI chains several actions into an end-to-end workflow.
The three terms overlap, and vendors use them loosely. This guide explains each in plain words, compares them with a work example and shows how to decide which one you need.
What is generative AI?
Generative AI creates new content, such as text, images or summaries, from a prompt you write. ChatGPT, Claude, Google Gemini and Microsoft Copilot are used this way when you ask them to draft an email or summarise a report. You drive every step: you write the prompt, read the result and decide what to do with it.
What is an AI agent?
An AI agent goes a step further. You give it a task and a set of rules, and it carries out several steps itself, using the tools you connect. For example, an inbox agent can read your messages, sort them by urgency, draft replies to routine queries and ask you about the rest. It checks back with you when a decision is needed.
What is agentic AI?
Agentic AI describes systems that plan and chain actions together to reach a goal, often across several tools and with less step-by-step prompting. In practice, it is a workflow where an agent, or several, handles a process from start to finish, while a person sets the limits and approves the key steps.
How do generative AI, AI agents and agentic AI compare?
| Generative AI | AI agent | Agentic AI | |
|---|---|---|---|
| What it does | Creates content from a prompt | Carries out a multi-step task using tools | Chains actions into an end-to-end workflow |
| Who drives each step | You prompt at every step | You set the task and rules, and it runs the steps | You set the goal and checkpoints, and it plans and runs the steps |
| Inbox example | Drafts a reply when you paste in an email | Sorts the inbox, drafts replies to routine queries and flags urgent ones | Reads incoming mail, updates the CRM, creates the follow-up task and sends you a summary |
| Your role | Review each output | Approve or correct its work | Set limits, review exceptions and approve key steps |
Which one do you need?
Match the tool to the job:
- You need a draft, a summary or an idea. Use generative AI. A good prompt is enough. Our guide to ChatGPT prompts for busy professionals shows how.
- The same steps repeat every day or week. Build an agent. Routine inbox work, invoice drafting and enquiry follow-up are common first choices.
- A process crosses several systems. Look at an agentic workflow, once a single agent works reliably.
Most teams begin with generative AI, move to one agent on a task with clear rules and grow from there. Our guide on what to automate with AI helps you choose that first task.
How do agents show up in real work?
At QD Academy, a follow-up agent answers people who register through a landing page form on WhatsApp straight away. When one of our team joins the chat, the bot goes to sleep so a person can carry on. We found that slow manual follow-up was losing leads, and after we automated the first reply, the response rate went up by 50% and the enrolment rate rose as well. Our guide to customer journey automation by stage describes the set-up, and 5 AI agents a Singapore SME can build in a day shows four more examples.
What are the risks and limits?
- Errors. AI can be wrong or confident about things it has made up. Review anything that goes to a customer or moves money.
- Access. Give an agent only the data and tools the task needs.
- Personal data. Check what happens to the data you send to an AI tool. Our PDPA checklist for using AI on customer data is practical guidance and is not legal advice.
- Ownership. Name one person who is accountable for what the agent does.
Frequently asked questions
What is the difference between generative AI and an AI agent?
Generative AI creates content when you prompt it, one step at a time. An AI agent carries out a multi-step task using tools, following rules you set, and checks with you when it needs a decision.
Is ChatGPT an AI agent?
ChatGPT is mainly used as a generative AI assistant: you prompt it and it replies. Some versions and connected tools can also take actions for you, which is where it starts to behave like an agent. Features change often, so check what your version can do.
What is agentic AI in simple terms?
Agentic AI is AI set up to pursue a goal by planning and chaining several actions, often across different tools. A person still sets the limits and approves the important steps.
Do I need to code to build an AI agent?
Not for the work agents we teach. Our AI-Powered Productivity course is built for professionals with no coding or AI experience, and learners build an agent for a real process from their own work.
Are AI agents safe to use with company data?
It depends on the tool and its settings. Check what data the tool stores and who can see it, limit what the agent can access and keep a person approving sensitive actions.
Where does this guide fit in our AI productivity series?
It is part of our series on using AI at work. Start with the complete guide to improving productivity at work with AI for the full process.
Learn to build an AI agent in one day
The first module of QD Academy’s one-day AI-Powered Productivity: Strategies for the Modern Workplace course covers the differences between generative AI, AI agents and agentic AI and where each fits into your day. You then compare ChatGPT, Claude, Google Gemini and Microsoft Copilot, and spend most of the day building and testing an agent for a real process from your work. No coding or AI experience is needed, and you bring your own laptop. Check the course page for the latest dates, fees and funding.