1. Diagnose your level
Take the assessment first. Do not start with advanced workflow design if you still struggle to write clear prompts.
Start here
How to use this manual properly
Most people stay stuck because they read about AI without deliberately practising it. This manual is designed to teach by doing: each lesson has an outcome, the reasoning behind the method, a prompt, a critique loop, and a practical drill.
2. Work in sequence
Each level builds on the last. Prompt clarity comes before verification. Verification comes before automation.
3. Practise immediately
After each lesson, apply the technique to a real task from work, study, business, or home life within 10 minutes.
The learning loop
- Read: understand the method and why it works.
- Use: copy the prompt or framework into your tool of choice.
- Inspect: look for gaps, assumptions, weak logic, and missing context.
- Improve: refine the prompt and compare outputs.
- Save: keep your best instructions as reusable templates.
What separates beginners from experts
- Beginners ask for answers.
- Operators ask for structured outputs.
- Advanced users design workflows and review loops.
- Experts build systems, choose tools intelligently, and know when not to trust AI.
Skill assessment
Find your starting point
This quick diagnostic prevents wasted time. It estimates whether you should begin at Beginner, Operator, Advanced, or Expert. Choose the option that sounds most like your current ability, not your ambition.
Learning roadmap
The curriculum path from beginner to expert
The roadmap is organised by capability, not hype. The goal is to build skill in the right order: clarity, structure, evaluation, reuse, orchestration, then judgment.
Level 1Beginner
Get useful results consistently
Understand what AI is good at, how to ask well, and how to avoid the most common beginner mistakes.
- Clear prompting and context
- Summaries, explanations, and drafting
- Basic verification and source awareness
- Prompt formatting and iteration
Level 2Operator
Use AI to do meaningful work
Apply AI to writing, research, analysis, meetings, files, and document-heavy tasks with repeatable structure.
- Research with checks and constraints
- Document extraction and transformation
- Meeting notes, action plans, and summaries
- Business-ready outputs
Level 3Advanced
Design reusable systems
Build strong prompt frameworks, use long context intelligently, and design tasks as multi-step workflows.
- Reusable prompt libraries
- Project context and long-document workflows
- Image and multimodal use
- Quality control and failure analysis
Level 4Expert
Orchestrate tools with judgment
Move from one-off prompts to systems that combine tools, human review, safety, and operational discipline.
- Agent-like workflows and automation logic
- Governance, privacy, and risk control
- Capstone design and measurement
- Tool selection by task, not brand loyalty
Interactive curriculum
The course-book lesson library
Filter by level or tool, search by topic, then study each lesson in full. Use the “mark complete” button to save progress.
Prompt lab
Frameworks you can reuse every day
Good users stop reinventing prompts. Strong operators create frameworks that can be reused, adapted, and shared. These are built to force clarity, control output shape, and reduce wasted rounds.
Tool selection
How to choose the right AI tool for the job
Experts do not pick tools based on tribal loyalty. They choose based on the task, required fidelity, available context, file handling, workflow needs, and how much checking the outcome requires.
| Task Type | What matters most | Good fit | Why | Caution |
|---|---|---|---|---|
| Learning and tutoringExplain a concept, test yourself, ask follow-up questions. | Clarity, structure, patience, guided interaction | ChatGPT | Works well for iterative teaching, guided practice, and project-style ongoing work. | Do not confuse fluent teaching with factual certainty. |
| Big document packsRead long reports, compare files, extract patterns. | Long context, document grounding, cross-reference ability | Gemini | Strong for long-context workflows and large document sets. | Long inputs still need structure, indexing, and clear extraction instructions. |
| Standalone outputsGenerate tools, rich content, interactive pieces. | Output packaging, artefact-style creation, iterative making | Claude | Very good when you want a substantial piece of content to work on separately. | Still review factual claims and edge cases. |
| Research and comparisonCompare options, summarise findings, produce a brief. | Source quality, structure, verification, scope control | All can work | The workflow matters more than the brand: ask for criteria, evidence, gaps, and uncertainties. | Research without checks becomes polished nonsense very quickly. |
| Writing and rewritingEmails, reports, speeches, messaging, summaries. | Tone control, audience fit, editing flexibility | All can work | Use the tool you can iterate with fastest, then inspect for voice, logic, and accuracy. | Never send sensitive or high-stakes writing without human review. |
Capstone studio
Projects that prove mastery
Each capstone requires you to combine prompting, structuring, checking, and judgment. Completion of these means you are not merely experimenting with AI — you are using it deliberately.
Glossary
The terms that matter, explained plainly
Jargon blocks learning. These definitions are written for learners who want practical meaning, not performative complexity.
Official resources
Primary references worth studying
These are useful official sources for learning core ideas behind prompting, projects, long context, artifacts, and major platform capabilities. Use them to extend what you learn here.
OpenAI — Prompting and projects
Read how clear prompting, iterative refinement, and projects help structure work across chats and files.
Anthropic — Artifacts and computer use
Explore how substantial outputs can be built in dedicated workspaces and where tool-driven computer workflows fit.
Google — Gemini long context
Review long-context guidance, model documentation, and structured generation patterns for large inputs and outputs.
How to study this material
Do not try to memorise tools feature by feature. Instead, study: task definition, context, verification, reuse, and workflow design.
Rule: A better workflow beats a flashier prompt.
Rule: A checked answer beats an elegant hallucination.
Rule: Reusable instructions beat repeated reinvention.
FAQ
Questions learners ask when they get stuck
These are the predictable friction points. Good design anticipates them and removes excuses early.
Reality check
Common misconceptions that hold people back
These are beliefs that sound reasonable but lead to poor decisions. Each one is common, confident, and wrong.
Quality calibration
What good actually looks like
Most people cannot tell the difference between acceptable and excellent AI output. These benchmarks calibrate your judgment.