Level 1Starting out
Recommended for you
What is AI? A plain-English starter
A jargon-free introduction to modern AI: how chatbots work, what they are good and bad at, and how to use them safely from day one.
Your path
Free AI course downloads
Learn AI.Free.Yours to keep.
The current free library takes you from your first prompt to inference engines and frontier models, with no login, card or account. Twenty additional premium courses are planned and are not open for purchase. Following Mickai helps us keep making the free workbooks.
- No account
- PDF workbooks
- Q&A at the end of every course
- Beginner to frontier
4 levels. Your own pace.How it works
- Free workbooks
- 107
- Levels, beginner to frontier
- 4
- Hands-on exercises
- 751
- Questions answered
- 1140
01How it works
How it works
-
Step 1: Choose your course
Explore the courses by subject, level, goal or industry. New to AI? The optional two-minute quiz can suggest a starting point.
-
Step 2: Study online at your own pace
Read the lessons, try the worked examples and complete the exercises in your browser. Download the PDF workbook when you want an offline copy.
-
Step 3: Check what you have learned
Use the chapter questions to review your understanding. Where a scored exam is available, pass with strictly above 80 percent for a named Mickai educational certificate. This is not professional accreditation.
The agent loop
Intelligence,
in motion.
From a goal to a checked result. Explore the four steps an agent can repeat, then see when it should stop.
Drag the ring to explore. Hold to dwell. Use arrow keys to change step, Home for Observe and End for Verify. Space pauses or resumes.
A result is a reason to check, not a reason to trust.
01 / 04Observe
Start with
the evidence.
Reads the goal, the conversation and the latest results.
An illustrated example
Find the museum's opening time.
What this step gives you
The goal, conversation and available information.
Read the four steps
Observe
Reads the goal, the conversation and the latest results.
Think
Reasons about the next step and chooses one action.
Act
Uses a tool: searches, writes a file or calls an API.
Verify
Checks the result against the goal, then loops or stops.
Go deeper in the free workbooks: What is an AI agent? Agentic AI explained then The loop harness: how AI agents actually run.
03The levels
Four levels. Start wherever you honestly are.
Level 1
Starting out
Everyone starts here, and it is a great place to start.
17 workbooks
Level 2
Getting going
You have the basics. Time to get properly good.
28 workbooks
Level 3
Building
You are ready to build things that work.
43 workbooks
Level 4
Frontier
You are at the edge. Let's go deeper.
19 workbooks
Level 1 of 4
Starting out
You will begin with plain-English foundations and gentle, hands-on exercises. No jargon, no coding.
- What is AI? A plain-English starter75 min
- Your first prompts: how to ask AI for what you actually want120 min
- The free AI toolkit: how to choose tools and stay safe120 min
- Make YouTube videos with AI: a beginner's workflow140 min
- Stay safe with AI: scams, fake media and your privacy100 min
- Use AI to study, research and write without losing your own thinking110 min
- Using AI to find a job: your CV, cover letter and interview prep70 min
- Using AI for everyday admin: letters, emails and forms55 min
- Using AI as an accessibility tool: reading, writing and seeing55 min
- Keep your job: use AI to do it better95 min
- A second income with AI: what actually works120 min
- Deepfakes and misinformation: how to tell what is real90 min
- Which AI model should I use right now? A living, refreshed comparison80 min
- Why does AI make things up? Hallucinations explained85 min
- Free AI tools vs paid: what you actually need to pay for85 min
- Writing better emails, reports and documents with AI70 min
- Spotting an AI voice-cloning scam70 min
Level 2 of 4
Getting going
You will learn to pick the right tools, build simple things and understand what agents are.
- AI images and video: rights, provenance, labels and quality checks105 min
- Build your first app with an AI coding assistant120 min
- Deploy AI agents without losing control95 min
- Git for AI builders: commits, branches and pull requests125 min
- HTML: build a meaningful, accessible page130 min
- Running open-weight AI models on your own computer, for free65 min
- Automate a spreadsheet workflow with AI, checks and an audit trail120 min
- Build and launch a website with AI120 min
- Legitimate voice cloning: how it works and the honest, consented uses75 min
- Prompt like the highest-paid engineer in the room165 min
- What is sovereign AI, and why it matters to you90 min
- AI skills employers actually want, and how to show them75 min
- CSS: responsive layouts, grids and typography140 min
- What is an AI agent? Agentic AI explained75 min
- AGI explained: what it means and what it does not75 min
- Agents vs chatbots: what's actually different55 min
- Editing and proofreading your manuscript with AI80 min
- Python: from first script to a useful automation140 min
- Writing fiction with AI: what actually helps85 min
- Code for as little as possible: VS Code, OpenCode and app builders125 min
- Vibe coding: what it is, what it's good for, and where it breaks90 min
- AI compliance for a small business, in plain English90 min
- AI dubbing and localisation for video: what it actually does90 min
- Self-publishing on KDP: the AI disclosure rules65 min
- AI voice scam prevention: verify callers and protect payments720 min
- AI literacy for school staff720 min
- AI literacy and task selection for public servants720 min
- AI literacy and authority in noncombat defence support720 min
Level 3 of 4
Building
You will learn how agents really run, how to build and host a chatbot, and how retrieval and knowledge bases work.
- AI Investment and Readiness for Leaders202 min
- The loop harness: how AI agents actually run90 min
- Build your first chatbot105 min
- Design a repeatable evaluation suite for an AI application95 min
- Document pipelines: parsing, cleaning and chunking120 min
- Tokenisation: train and inspect a small tokenizer105 min
- Embeddings and vector search from first principles95 min
- Host a chatbot on your website, safely90 min
- Prompt injection: threat analysis and defensive testing135 min
- SQL: model and query your first database140 min
- Tensors, gradients and a first training loop95 min
- Test-driven development with an AI coding assistant150 min
- Data analysis with Python: investigate and explain a dataset90 min
- GPU memory maths for LLMs and SLMs75 min
- JavaScript: the browser, events and application state140 min
- Node.js: build and test a small API150 min
- Privacy and data minimisation in AI applications100 min
- RAG and knowledge bases, step by step100 min
- Tool calling and structured outputs: validate every response95 min
- Enterprise on-prem AI adoption: a practical guide130 min
- Go: build a concurrent service180 min
- Open model licences, datasets and deployment obligations110 min
- Setting up a workstation or home server to run AI locally90 min
- TypeScript: types for safer web applications180 min
- AI governance: risk registers, evidence and human accountability115 min
- Testing agent loops: budgets, termination and failure recovery150 min
- Building an ISMS for an AI-enabled organisation250 min
- AI prospect research without invented intelligence720 min
- AI product catalogues and truthful descriptions720 min
- AI-assisted legal intake and conflict screening720 min
- Digital evidence: AI provenance and custody records840 min
- AI-assisted vocational lessons and demonstrations720 min
- AI product briefs grounded in user needs and accountability720 min
- AI market research and defensible claims720 min
- Controlling payment proposals from financial AI agents780 min
- Digitising telecoms network and service records for controlled AI780 min
- Law-firm knowledge retrieval across matter boundaries720 min
- AI governance and service accountability for telecoms providers720 min
- Digitising healthcare records for confidential AI retrieval780 min
- Engineering AI purpose, hazards and professional sign-off720 min
- Defence stores records for noncombat inventory720 min
- Law-firm AI governance and supplier assurance720 min
- Accessible AI assistants for citizen enquiries720 min
Level 4 of 4
Frontier
You will go inside inference engines and reasoning models, and learn how to judge a new frontier model the day it lands.
- Inference engines explained80 min
- Reasoning models and reasoning engines80 min
- How to evaluate a new frontier model the day it lands150 min
- Hybrid search and reranking for better retrieval180 min
- Throughput, latency and batching: capacity planning for a model server180 min
- Transformers: attention explained and implemented180 min
- Build a tiny language model from scratch180 min
- Evaluate a RAG system with a grounded test set180 min
- Model routing, caching and graceful fallbacks180 min
- Permission-aware retrieval and private knowledge bases180 min
- Rust: ownership, safety and a small command-line tool180 min
- Harness state and agent memory: checkpoints, retention and deletion180 min
- Knowledge graphs and structured retrieval180 min
- Threat-model an agent and its tool boundary180 min
- Orchestrating multiple agents with explicit responsibilities180 min
- Agent protocols and context engineering230 min
- Running AI Agents Safely in Production220 min
- Human-in-the-loop vs human-in-time: oversight before impact600 min
- Healthcare AI purpose, governance and clinical safety boundaries780 min
04The library
Every workbook is a free PDF
Level 1Starting out
Recommended for you
Your first prompts: how to ask AI for what you actually want
Learn to ask AI for what you actually want: the six parts of a good prompt, how to improve a reply, and how to build a personal prompt library. Free tools only.
Level 2Getting going
Recommended for you
AI images and video: rights, provenance, labels and quality checks
Use AI images and video responsibly: ownership in the UK, EU and US, tool terms, consent, labelling, Content Credentials, quality checks and simple records.
Level 2Getting going
Recommended for you
Build your first app with an AI coding assistant
Build a small bill-splitting web page with any AI coding assistant while staying in control: brief it, build in steps, review the code, test edge cases and keep safe copies.
Level 3Building
Recommended for you
AI Investment and Readiness for Leaders
A practical method for leaders investing in AI: define the problem, compare the alternatives, build an honest business case, check readiness, then pilot it inside real gates.
Level 4Frontier
Recommended for you
Inference engines explained
How inference engines run a model: memory maths, quantisation, prefill and decode, batching, hardware, and serving a model safely over a local API.
05Why free
Why we are giving this away
We went to the depths of what is possible with AI today, and we learned a great deal building Mickai. That knowledge should not be withheld. So we are giving it to the public, for free, so anyone can take it and go and build something of their own.
On the purpose of Trust Agent
“free training to people who cannot afford to pay”
Mickai is not an open source project. We are building a consolidated ecosystem of software platforms, an operating system for organisations that want to own their intelligence. That work took the team a long way into what AI can and cannot do.
Along the way we learned things that are useful far beyond our own product: how agents really run, how to build a chatbot that behaves, how retrieval works, how inference engines are put together, how to judge a new model when one lands.
That knowledge should not be locked away. These workbooks are our attempt to hand it over in a form anyone can use, so people can take it and build projects of their own, and move AI forward in ways we will never think of ourselves.
06The team
The people behind Mickai
Trust Agent is published by Mickai LTD. These are the people behind Mickai, with the roles and profiles published on the Mickai website.
Abbie Youdale
Managing Director
Leads Mickai as Managing Director, with overall responsibility for how the company is directed and run. She oversees the executive team across leadership, compliance, governance and product development, sets the company's priorities and holds every function to account for delivery.
Mickarle Wagstaff-Irons - Micky Irons
Founder · CEO · Inventor
Named inventor on the full portfolio of 104 filed UK patent applications, owned by Mickai LTD. Original co-founder of Collector Crypt, now a live on-chain marketplace with more than $1 billion in transaction volume. Engineering background in nuclear at Sellafield and fusion at the UK Atomic Energy Authority.
Mickarle Wagstaff-Irons - Micky Irons on LinkedIn
Susan Brown
Chief Compliance Officer · Director
Leads Mickai's compliance function and contributes board-level oversight across technology, agreements and operational delivery. Her remit includes ISO/IEC 27001 readiness, Cyber Essentials preparation, regulatory alignment and evidence-led software release governance. Founder and Chairwoman of Zortrex, Director of Company Policy, with more than 40 years of bookkeeping and accountancy experience.
Susan Brown on LinkedIn
Niall Ogilvy
Chief Governance Officer
Governance, delivery and assurance leader with more than 25 years across 11 sectors. A Royal Navy and British Army veteran with experience in defence, the NHS, transport and manufacturing, he leads clear ownership, accountability and measurable progress.
Matthew Chubb
Chief Technology Officer
Hands-on software engineering, data and AI leader who has built language models since 2017 and taken products from proof of concept to production. Leads the engineering discipline behind Mickai's Sovereign Intelligence Operating System, with a focus on customer-led design, maintainable software, rigorous testing and dependable operation beyond a demonstration.
Matthew Chubb on LinkedIn
Vivek Shah
Head of Growth & Capital Formation
Commercial operator and capital-formation specialist with more than a decade of experience building revenue engines, go-to-market functions and investor relationships across venture-backed technology businesses. Has built commercial teams from the ground up, generated more than $1 million in ARR, and led fundraising across angel, family-office and institutional investor networks. At Mickai, leads go-to-market strategy, revenue, strategic partnerships, investor relations and fundraising as the company moves into commercial scale.
Vivek Shah on LinkedIn
Chris Williams
Chief Business Development Officer
Technology executive in defence, AI, and blockchain, with a record of mission-critical systems and government contracts. Owns the MOD, Cabinet Office, NCSC, AISI, and Five Eyes pipeline and the sovereign-AI procurement strategy.
Chris Williams on LinkedInMickarle Wagstaff-Irons - Micky Irons, full name Mickarle Sean Junior Wagstaff-Irons. Profiles and original portraits reproduced from the Mickai website.
07Questions
Questions, answered
Anything else? Write to hello@mickai.co.uk.
Which courses are free?
Every published workbook in the current free library is a free PDF download, with no payment or trial. Twenty additional premium courses are planned and are not open for purchase. Proposed referral credits cannot currently be earned or spent.
Do I need to sign up or give an email address?
You do not need an account. Course registration, download and exam forms ask for your name and email and a privacy acknowledgement. Newsletter signup is a separate optional choice, confirmed by email. Following Mickai on LinkedIn is optional and is not verified by this site.
What is a workbook?
A workbook is a PDF course you work through at your own pace. It has plain-English explanations, hands-on exercises with space to write, a checklist, a glossary and a Q&A at the end so you can test yourself.
Who writes the workbooks?
The workbooks are written by Mickarle Sean Junior Wagstaff-Irons - Micky Irons and published by Mickai LTD. They are drafted with AI assistance and checked against primary sources. Review methods and limitations are recorded internally; a review date is not a claim of independent human review.
Why is Mickai giving these away?
Because the team learned a great deal building Mickai, and that knowledge is more useful shared than withheld. Mickai is not open source; this is a gift of knowledge so that others can build their own projects.
I have never used AI. Where do I start?
Take the quiz and choose that you are new to AI. You will be started on What is AI, then your first prompts. No jargon and no coding required.
I already build with AI. Is there anything for me?
Yes. The Building and Frontier levels cover AI agents and loop harnesses, building and hosting a chatbot, retrieval augmented generation and knowledge bases, inference engines, reasoning models, and how to evaluate a new frontier model.
Can I share the workbooks?
You are welcome to use them personally, in study groups and inside your own team, and to point people to trust-agent.ai. Please do not resell them or republish them as your own.
How are downloads counted?
We count download events using course, date, coarse country and a daily pseudonymous request fingerprint. These are not verified unique people. The separate download form privately records name, email and request IP with a privacy acknowledgement. See the privacy notice for retention.
Free AI course downloads