Ghost — Guide

Your First Week With AI

A day-by-day walkthrough that turns AI from something you tried once into something you actually use. No coding. No jargon. Seven days of concrete steps.

Ghost · May 18, 2026 · 7-day series · Practical guide

Most people try AI three times before giving up. First time: they ask it something vague, get a generic answer, and decide it's overhyped. Second time: they try a different tool, ask a slightly better question, get a slightly better answer, but still don't know what to do with it. Third time: they open ChatGPT, stare at the blank box, close it, and don't come back.

This pattern is not your fault. The tools present themselves as magic, but what they actually require is a skill nobody teaches. The skill is not "prompt engineering." It is: knowing what to ask, how to give it the information it needs, and what to do when the answer is not quite right.

This walkthrough teaches that skill in seven days. Each day has one task, takes fifteen to thirty minutes, and builds on the day before. By day seven, you will not be an expert. You will be someone who has turned AI from a curiosity into a tool with at least one job it does regularly for you.

Before you start: pick one tool

There are hundreds of AI tools. You will use exactly one this week. If you don't have a preference, use ChatGPT at chat.openai.com. Claude (claude.ai) and Gemini (gemini.google.com) are equally good alternatives. Pick one and ignore everything else for the next seven days.

The tool itself is not the constraint on your results this week. The constraint is whether you learn how to use any tool well. Switching tools before you learn the skill is the fastest way to stay at day zero.

The Seven Days

Each day is a standalone task. You can do them in order or jump around, but the skills build. Start with Day 1 and see how far you get.

Day 1
Your First Task
Pick something small you already wrote and improve it with AI. The goal is one genuinely better piece of work.
Day 2
The Example Lever
Give AI an example of what "good" looks like. Single highest-leverage quality improvement in the entire series.
Day 3
Iterate, Don't Accept
Every AI output is a draft. Learn to tell it what to change instead of starting over.
Day 4
Organise Your Thinking
Use AI to find patterns in your own notes and raw thoughts. A different type of task that plays to AI's real strengths.
Day 5
Find Your Hook
The most important day. Identify one recurring task AI can do for you every week and create a saved prompt for it.
Day 6
The Trust Check
Three sixty-second checks that catch most AI errors before they cause problems.
Day 7
Your First Review
Take stock, diagnose what went wrong, and lock in your gains with the one-in-one-out rule.

What This Week Is Really About

This walkthrough does not teach "prompt engineering." It teaches something more fundamental: how to turn a general-purpose tool into a specific-purpose tool by giving it the right constraints.

The skill you will practise every day — providing context, showing examples, iterating on output — transfers to any AI tool you will ever use. It also transfers to working with people: clearer instructions, better examples, more specific feedback. The AI is just a patient practice partner.

The difference between someone who gets value from AI and someone who does not is rarely about which model they use. It is about whether they have built the habit of treating AI output as a starting point rather than an end point, and whether they have embedded at least one recurring task into a routine. That is what this week builds.

Start with Day 1 Day 1 →
Subscribe for More

The Durability Curve on Substack

Practical guides, earnings analysis, and framework essays delivered to your inbox. Free weekly dispatches from the research pipeline.

Subscribe Free →