What is an AI agent?

An AI agent is software given one defined job, access to the systems it needs to do that job, and written instructions on how the work should be done — so the job gets carried out repeatedly without a person doing it each time. If a business is a team, an agent is the member of staff who owns one role.

The useful way to think about an agent is as a hire. You decide what the role is, you give them access to the systems the role needs, and you brief them on how you want it done. The difference is that the briefing is written down, so it never drifts, never gets forgotten, and never leaves.

What separates an agent from a chatbot is that an agent has a job rather than a conversation. A chatbot waits to be asked something; an agent reads the overnight enquiries, drafts the replies, files what needs filing, and flags the one that looks wrong. Built properly it drafts and flags — a person still sends anything that reaches a customer.

One agent per job beats one agent for everything, for the same reason you would not ask your bookkeeper to write your marketing. A narrow role with a clear brief produces better, more predictable work than a general-purpose assistant asked to do a bit of all of it.

The service: SilentShift does this work — AI agentic integration, from £2,500.

Going deeper: Agents, skills and identities in Contextium — the platform SilentShift builds them in (a declared interest: see the About page).

← All glossary terms