Building Your First AI Agent — book cover

Book 6 of 10 · Intermediate Operator tier

Building Your First AI Agent

Stop drafting. Start delegating — to an agent that does the job.

+ add the Advanced Vault pack for £27.00 at checkout

Not sure this is your book? Take the 60-second quiz →

A chatbot writes you a draft. An agent actually does the work: it takes an instruction, uses tools, and finishes the task. This book walks you through building your first AI agent on no-code platforms — safely, with a human in the loop and guardrails in place — so you can hand over a real job, not just a first draft.

It's deliberately careful about sequencing: understanding what an agent actually is and isn't, giving it a clear brief and the tools and memory it needs, and — critically — choosing a genuinely safe first job before building anything. Only then does it move to building and testing on no-code agent platforms.

The second half is about keeping an agent trustworthy once it's live: a human in the loop, guardrails and access limits, logging, and a full chapter on what to do when an agent goes wrong — because it will, eventually, and the book doesn't pretend otherwise.

Who it's for

People comfortable with AI day-to-day who are ready to move from asking it questions to delegating it a job.

What you'll learn

  • Understand what an agent really is and pick a safe first job for one
  • Give an agent a clear brief, the right tools and the memory it needs
  • Build, test and launch your first agent on no-code platforms
  • Keep a human in the loop with guardrails, access limits and logging
  • Know the warning signs and when an agent is the wrong answer

Inside the book

  1. 00From Drafting to Doing
  2. 01What an Agent Actually Is
  3. 02The Agent's Brief
  4. 03Tools: How an Agent Acts
  5. 04Memory & Knowledge
  6. 05Choosing a Safe First Job
  7. 06The No-Code Agent Platforms
  8. 07Building Your First Agent
  9. 08Testing an Agent
  10. 09Human-in-the-Loop
  11. 10Guardrails, Access & Logging
  12. 11Watching It Work
  13. 12When Agents Go Wrong
  14. 13Cost & When Not to Use an Agent
  15. 14Your 30-Day Agent Plan

What this can look like at this level

The kind of week each book is built to produce.

A veterinary or healthcare practice

clinical notes eating into time between appointments. Moved to ambient note capture with a structured summary the practitioner reviews.

Practices commonly report this kind of tool meaningfully cutting note-writing time, freeing up time for direct patient care.

A professional services firm or agency

proposals and tender responses rewritten from scratch each time. Moved to a reusable knowledge base with AI-drafted first-pass sections.

This usually shortens proposal turnaround and makes it easier to respond to more opportunities.

Questions people ask about this book

How is an agent different from the automations in Book 3?

An automation follows a fixed set of steps you defined. An agent is given a goal and tools, and works out the steps itself — Chapter 1 draws this distinction in full before anything gets built.

Is this safe to hand real business tasks to?

The book's whole structure is built around that concern — choosing a genuinely safe first job (Chapter 6), keeping a human in the loop (Chapter 10), and guardrails and logging (Chapter 11) all come before anything is called "done."

Do I need to know how to code to build an agent?

No — it uses no-code agent platforms throughout, the same click-together style as Book 3's automations.