AI glossary for businesses
AI agent
In short
An AI agent is an AI system that works towards a goal by planning steps, using tools and acting on its own, rather than only answering one question at a time. It can read emails, look up data, update a spreadsheet or draft a reply, then check the result and choose the next step, within limits you set.
How an AI agent works
A chatbot is like a knowledgeable colleague you ask questions. An agent is more like an intern you give a task: "collect this week's supplier delays and update the delivery sheet". The intern decides the steps, opens the right systems, does the work and reports back.
Under the hood, an agent combines three things:
A language model that reads, reasons and decides the next step.
Tools it is allowed to use, such as email, a calendar, a CRM, web search or a spreadsheet.
A loop: plan, act, check the result, repeat until the goal is reached or it needs a human.
Products such as ChatGPT's agent features, Microsoft Copilot Studio and the AI features in Zapier and Make let you build simple agents without heavy coding.
In practice for a small business
A small distributor receives dozens of supplier emails a week about changed delivery dates. Today someone reads each one and updates a shared spreadsheet. An agent could:
read new emails in a shared inbox;
pick out the order number and new date;
update the matching row in the spreadsheet;
draft a note to the affected customer and leave it for a person to approve.
Start with a narrow, repetitive task like this, keep a human approval step and run it as a pilot alongside the manual process for a few weeks. The article AI agents for SMEs explained covers more options.
Why it matters and what to watch out for
Errors add up: a small mistake in step one can carry through every later step. Test with real cases and log every action.
Least access: give the agent only the permissions it needs. Read-only where possible. No payments or deletions without human sign-off.
Personal data: agents often touch emails and customer records, so GDPR applies. Map what data it handles; at scale, consider a DPIA.
High-risk uses: under the EU AI Act, using AI to screen job candidates is a high-risk use case, with obligations phasing in through 2026–2027. Check with your DPO or legal advisor before such uses.
Not always needed: if the steps never change, simple workflow automation may be cheaper and more reliable.