AI Agent.

What is an AI agent?

An AI agent is software that uses a large language model to pursue a goal by deciding its own next steps, using tools such as email, a CRM or a web search, and checking its results as it goes.

A chatbot answers a question. An agent is given a job, such as 'qualify this enquiry and book a call if it fits', and works out how to complete it within the limits you set.

Why it matters

Agents can take on the judgement-light admin that eats a team's week: sorting enquiries, chasing missing paperwork, preparing quotes from notes, updating records. For a firm where the owner still does that work at 9pm, an agent that handles the first pass is time back. The flip side is that an agent acts on your behalf. A badly scoped one can email the wrong client, change the wrong record or spend money, so the design matters as much as the model.

How it works

• A clear goal You give the agent an outcome to reach and the rules it must follow on the way.
• A reasoning loop The model plans a step, takes it, looks at the result and decides what to do next until the goal is met or it gets stuck.
• Tools and connections Connections to your systems, such as the inbox, calendar or CRM, let it read and act rather than only write text.
• Context and memory It works from the information you supply: your policies, past records and the task history so far.
• Guardrails and approvals Permissions, spending limits and approval steps decide what it may do alone and what needs a person.

What to do about it

Pick one repetitive task that follows roughly the same steps each time, such as triaging new enquiries. Write down exactly what a good member of staff does, including the exceptions. Decide what the agent may read, what it may change and what must go to a person for approval. Run it alongside a human for a few weeks, compare the results, and only then let it act alone on the low-risk parts.