AI, put plainly.
Subscribe free →
1
Understand
2
Identify
3
Prompt
4
Refine
5
Build
1
Stage one
Understand what AI is actually doing

Before you can use AI well, you need a mental model for what it actually is — not the sci-fi version, not the hype version. AI is a pattern-completion engine. It has processed an enormous amount of human text, and when you give it instructions, it produces the most useful response it can based on what it's learned.

That means it's incredibly good at tasks that involve language: summarizing, drafting, explaining, restructuring, translating ideas. It's less reliable when it needs to verify real-time facts or perform complex calculations without help.

The good news: you don't need to understand any of the technology underneath. You just need to understand the relationship — you're the expert on your situation, AI is the expert on language. Together you're more powerful than either alone.

Key insight: AI doesn't "know" things the way you do. It produces responses based on patterns. Your job is to give it enough context that those patterns produce something genuinely useful for you.

Real-world examples
A law student uses AI not to cite cases — AI can hallucinate those — but to summarize dense readings, explain concepts three different ways, and test their understanding before an exam. The student brings the legal judgment; AI brings the language.
A small business owner uses AI not to run their business — they know it better than any AI — but to draft emails, write product descriptions, and summarize customer feedback in a fraction of the time. They bring the knowledge; AI brings the speed.
A job seeker uses AI not to apply for jobs — only they can decide what's right — but to tailor cover letters, decode job descriptions, and practice interview answers on demand. They bring the ambition; AI brings the polish.
Stage 1 of 5
2
Stage two
Identify where AI fits in your work

Not every task benefits from AI equally. The fastest wins come from identifying work you do that is repetitive, language-heavy, and time-consuming — but doesn't require your unique judgment to initiate.

Think about tasks you do every week: drafting emails, summarizing documents, writing reports, explaining things to others, researching topics, structuring ideas. These are your starting points.

The goal at this stage is simple: make a list of five tasks in your work or daily life that fit this pattern. You don't need to change anything yet — just see where the opportunity is.

Key insight: The best first AI tasks are ones where a "pretty good" first draft saves you significant time — even if you edit it heavily afterward. Done is better than perfect when the alternative is starting from nothing.

Finding your best AI tasks
A law student identifies that making study guides from lecture notes takes two hours and follows a consistent format every time. That's a perfect AI task — consistent structure, language-heavy, doesn't require legal judgment to produce.
A small business owner identifies that responding to customer enquiries takes 30 minutes a day and covers the same ten questions repeatedly. AI can draft those replies in seconds — leaving more time for actual business decisions.
A job seeker identifies that tailoring their CV for each role takes an hour per application. AI can do a strong first draft in minutes — they just need to review and personalize it.
Stage 2 of 5
3
Stage three
Write prompts that actually work

A prompt is just an instruction. But the quality of your instruction determines the quality of the output. Most people write vague prompts and get vague results — then blame the AI.

The Plainly prompting method has four parts, remembered as RCTF: Role (tell AI who it's being), Context (give it relevant background), Task (say exactly what you need), Format (specify how you want the output). You don't always need all four — but the more you include, the better the result.

Key insight: Treat AI like a brilliant assistant on their first day. They're capable — but they need context. The more you explain your situation and what "good" looks like, the better they perform.

Try this prompt — copy and paste it into any AI tool
You are a [your role — e.g. law student / small business owner / marketing manager]. I need help with [describe your task]. Here is the relevant background: [paste any notes, documents, or context]. Please produce [describe the output — e.g. a 200-word summary / a bullet point list / a draft email]. Keep the tone [professional / friendly / concise].
Notice the structure: Role → Context → Task → Format. Fill in the brackets with your own details and you have a strong prompt for almost any situation.
Weak vs. strong prompts
Weak: "Write me a cover letter."

This produces a generic letter that could be for anyone, applying for anything. Completely unusable without major rewriting.
Strong: "You are a career coach helping a recent law graduate apply for a trainee solicitor role at a commercial law firm. Using the job description below and my CV summary, write a compelling 250-word cover letter that highlights my research experience and interest in corporate law. Keep the tone professional but not stiff."

This produces something you can almost send immediately.
Stage 3 of 5
4
Stage four
Refine through conversation

Most people try a prompt once, don't love the result, and give up. That's the wrong approach. The first output from AI is a starting point, not a final answer. The real power comes from the conversation that follows.

Think of it like editing with a collaborator. You can say: "Make this shorter." "Rewrite the second paragraph to be more direct." "Give me three alternative versions of the opening." "Now make it sound less formal." AI responds to all of these — immediately, without complaint.

The skill of refinement is learning how to diagnose what's wrong with an output and translate that into a clear instruction. This improves quickly with practice.

Key insight: A conversation with AI is like an editing session. Each message is a revision instruction. The quality of your final output depends on how well you guide the process — not just the first prompt.

A real refinement conversation
You: "Draft a short bio for my LinkedIn. I'm a second-year law student interested in tech law and I did a summer internship at a startup."

AI gives you a generic, formal bio that sounds like everyone else's.
You: "Good start — but make it sound less like a CV and more like a real person wrote it. Add a specific detail about why I'm interested in tech law."

AI revises — now it's warmer and more specific, but still a bit long.
You: "Cut it to 80 words and put the most interesting sentence first."

Result: Something you'd actually post. Three exchanges. Under five minutes. You directed every step — AI did the language work.
Stage 4 of 5
5
Stage five
Build a repeatable workflow

Once you've found prompts that work, save them. A prompt that produces great results once will produce great results every time you use it with similar inputs. This is how you go from "I used AI once" to "AI is part of how I work."

A workflow is a sequence: a trigger (something happens), an AI step (you apply your prompt), and an output (something useful is produced). The simplest workflows are just one step. More advanced ones chain multiple prompts together.

Start with one workflow from your Stage 2 list. Run it for two weeks. Refine it. Then add a second. Within a month, you'll have a personal AI system built around your actual life.

Key insight: The goal isn't to use AI for everything. It's to identify the 20% of your tasks where AI saves 80% of the effort — and systematize those. That's where the real gain lives.

Example workflows
Trigger: Lecture or reading session ends.
Step 1: Paste notes into saved prompt → AI creates a structured study guide with key terms and concepts.
Step 2: Ask AI to generate 10 practice questions from the material.
Result: Study materials ready in minutes instead of hours.
Trigger: New customer enquiry arrives.
Step 1: Paste the enquiry into saved prompt → AI drafts a warm, professional reply.
Step 2: Quick review and personalize → send.
Result: 20 minutes of daily email time cut to 5.
Trigger: Find a job posting you like.
Step 1: Paste posting + CV summary into saved prompt → AI tailors your cover letter to the role.
Step 2: Review, adjust the personal details → submit.
Result: Applications that took an hour now take 15 minutes.
Stage 5 of 5

You've completed The Framework.

You now have a mental model for AI, a method for prompting it, and a path to building workflows that save real time. The next step is putting it into practice — starting with one task, this week.