Course · 8 lessons · 60 min
Prompt engineering for daily work
A shortcut saves you seconds. A good prompt saves you a draft. This is the practical half of prompting — how to ask for the email, the spreadsheet formula, the post, the analysis — backed by a library of 57 prompts you can copy and keep.
Read it in order if you are starting out. Jump to whichever lesson matches the thing on your screen if you are not.
How prompting actually works
The mental model behind every good prompt — and the four parts that turn a vague question into a useful answer.
Context, constraints & giving good source material
Why pasting the real document beats describing it, how much context is too much, and the constraints worth naming every time.
Email, Slack & hard conversations
Prompts for the writing you do every day — replies, follow-ups, saying no, chasing people, and the message you are dreading.
Working with data, spreadsheets & documents
Extraction, cleanup, formulas and analysis — plus the arithmetic trap that quietly produces confident wrong numbers.
Writing posts, docs & content that isn't slop
Why AI writing reads as AI writing, and the prompting habits that fix it — angle, evidence, voice, and using the model as an editor instead of a ghostwriter.
Meta-prompting: let it write the prompt
The technique that makes every other technique optional — asking the model to build, diagnose and improve your prompts instead of engineering them yourself.
Removing AI patterns from text
The specific tells that mark writing as generated, why “make it sound human” never works, and what to say instead.
Advanced techniques & building your own system
Decomposition, forcing reasoning into the open, adversarial checks, reusable instructions — and the failure modes worth knowing before they bite.
Written for real work
Every example comes from something you actually do — a reply you are dreading, a messy export, a post with no angle.
Copy-paste templates
30 worked examples in the lessons, plus a library of 57 ready-to-use prompts. Swap the brackets and go.
Tool-agnostic
Claude, ChatGPT, Gemini — none of this depends on a menu that will move next month. It is about what you put on the page.
Unlike the shortcut pages on KeyMap, this course is not verified against a vendor’s documentation — there isn’t one to verify against. It is practice, not specification: what reliably works across current AI assistants, written to stay true as the tools change.