Edition 06 · Free PDF · Convert

Your First AI Employee

A non-technical guide to autonomous AI agents — what they are, what they can do, and the permission ladder that keeps them safe.

12 pages 12 min read A4 print-ready Free PDF

Get the free guide.

AI agents are not chatbots. They're software that works while you sleep — reading, deciding, drafting, and sending on a schedule or when an event happens. This 12-page guide explains what an AI agent actually is, what tasks it can handle, what it costs, and the permission ladder that lets you trust it one level at a time.

We'll email you the download link. No spam, no newsletter — just the guide.

What's inside

6 sections, 12 pages.

  1. Not a chatbot — a colleague on the night shiftA chatbot waits for you to type. An agent wakes up at 6am, reads your inbox, drafts three replies, and flags the one that needs your attention — before you've opened your laptop.
  2. What an agent can actually doTwelve candidate tasks, ranked by reliability. The top three work today with no custom code. The bottom three need tooling that doesn't exist yet. The guide is honest about both.
  3. The permission ladderLevel 1: read only. Level 2: draft for your approval. Level 3: send to approved recipients. Level 4: full autonomy. Most businesses should stay at level 2 for a month. The guide explains why.
  4. What it costs, honestly£3–40/month in model costs. The computer is one you already own. Setup is one afternoon. Weekly attention is 10–30 minutes. The guide puts this next to the loaded cost of the human task it replaces.
  5. The first agent to buildA daily inbox summary. Runs every weekday at 8am. Reads your unread emails, writes a three-line digest, sends it to your phone. The guide walks through the whole build in plain English.
  6. When to call for helpIf the task touches customer money, if it needs three or more systems connected, or if you can't explain what it does to a colleague — get help. Both choices are sensible.
Common questions

Questions this guide answers.

What are AI agents?
An AI agent is software that runs on a schedule or in response to an event, reads input from one or more sources, makes a decision based on instructions you've written, and produces output — without you opening a window and typing a question. The difference from a chatbot is that the agent acts while you're not looking. The guide explains the distinction with concrete examples from UK small businesses.
What can AI agents automate?
The most reliable agent tasks for UK SMEs are: reading and summarising incoming emails, drafting replies for human review, monitoring a webpage for changes and alerting you, compiling a weekly report from multiple data sources, and checking a calendar for conflicts. The guide lists twelve candidate tasks with honest notes on which ones are ready now and which ones need more tooling.
How much do AI agents cost?
Running cost is typically £3–40/month in AI model usage, depending on how often the agent runs and how complex the task is. The computer it runs on is usually one you already own (£0) or a small always-on cloud box (£4–10/month). Setup time is one afternoon. Weekly attention is 10–30 minutes reviewing what it did. The guide compares this to the loaded cost of the human task it replaces.
Are AI agents safe for business use?
They are as safe as the permissions you give them. The guide introduces a four-level permission ladder: level 1 is read-only (the agent reads but never writes), level 2 is draft-only (it writes a draft that you approve), level 3 is send-with-whitelist (it sends only to approved recipients), level 4 is full autonomy (it acts without review). Most businesses should start at level 2 and stay there for at least a month.

Read the guide. Then talk to us if you want to.

We'll tell you on the first call if the guide is all you need.

Book a chat