Switch AI
🧠 Level 4 — Context is Everything
Level 4 · Lesson 12 · 25 XP

Giving the AI your own material

⏱ 7 min ✦ 25 XP 🧠 3 questions
Reading
Demo
Exercise
Quiz
XP

Stop asking it to invent

Most beginners use AI as a generator — they ask it to write something from nothing. And it can do that. But the results are always generic, because the AI has nothing specific to work with.

The more powerful use is as a processor. You give it your material — your writing, your notes, your documents — and ask it to do something with what you've already got. The output is specific because the input is specific.

Your material + AI capabilities beats starting from scratch every time. The AI can summarise, rewrite, extract, restructure, analyse, translate, and transform. But it can only do those things well when it has something real to work with. That something is yours to provide.

This is the last piece of the context picture. Lesson 10 taught you how the AI's memory works. Lesson 11 taught you how to brief it about yourself. This lesson teaches you how to give it the raw material for the task at hand.

The formula

Paste-and-instruct is a pattern worth naming — because once you see it, you use it everywhere. It has two parts:

✦ The paste-and-instruct formula
Context block
"Here are my meeting notes from today: [paste notes]"
+
Instruction
"Turn these into action items with owner and deadline for each."
=
Useful output
A structured action list drawn from your actual notes — not a generic template.

The context block is what you paste. The instruction is what you want done with it. You're not asking the AI to invent — you're asking it to process something real. That distinction changes everything about the quality of the output.

Six things you can paste

The formula works across almost any type of content. Here are the most useful starting points:

📋
Meeting notes
"Turn these into action items with an owner and deadline for each."
✉️
An email
"Rewrite this to be more direct — remove the filler phrases."
📄
A document
"Summarise this in 3 bullet points for an exec who has 30 seconds."
✍️
Your own writing
"Write something new in this style — match my voice exactly."
📊
Data as text
"What patterns do you notice in this data? What stands out?"
💼
A job description
"What skills are they really looking for? What's the culture signal?"
Same material, completely different outputs. The context block stays the same. Only the instruction changes. One set of meeting notes can become an action list, a summary email, a project update, or a list of open questions — depending on what you ask. The AI doesn't decide what's useful. You do.
Interactive demo
Same meeting notes. Three different instructions. Watch what changes.
↓ Context block — pasted meeting notes
Team standup — Monday 14 April
Present: Sarah (PM), James (engineering), Priya (design), Marcus (marketing)

Sarah: Sprint review went well. Client happy with the dashboard changes. Two items carried over — the export function and the notification settings. James says export will be done by Wednesday. Notification settings pushed to next sprint, needs more scoping.

Priya: Onboarding flow mockups are 80% done. Needs feedback from Sarah before she can finalise. Hoping to have final version by end of week.

Marcus: Blog post about the new feature is drafted and ready for review. Needs a quote from Sarah. Also flagged that the help docs haven't been updated to reflect the dashboard changes — this is blocking customer support.

Next standup: Wednesday same time.
Instruction sent
"Summarise these meeting notes in 3 sentences for someone who wasn't there."
AI response
The sprint review went well with the client satisfied on dashboard changes, though two items — the export function and notification settings — carried over, with export due Wednesday and notifications deferred to next sprint. Priya's onboarding mockups are nearly complete and awaiting Sarah's feedback, while Marcus has a blog post ready pending a quote from Sarah. An outstanding issue is that help documentation hasn't been updated to reflect the dashboard changes, which is currently blocking customer support.
AI response
Action items:

1. Complete export function — James — by Wednesday
2. Scope notification settings for next sprint — James / Sarah — before sprint planning
3. Review onboarding flow mockups — Sarah — before end of week
4. Provide quote for blog post — Sarah — no deadline set
5. Update help docs to reflect dashboard changes — Owner unclear — urgent (blocking support)
AI response
Following our team standup on Monday 14 April, I wanted to share a brief update on current priorities.

The recent sprint review concluded positively, with the client expressing satisfaction regarding the dashboard updates. Two outstanding items remain in progress: the export function is expected to be completed by Wednesday, and notification settings have been scheduled for scoping in the next sprint.

On the design side, the onboarding flow mockups are nearing completion and are currently under review. In marketing, a feature announcement is drafted and awaiting final approval. Please note that an update to the help documentation is required as a matter of priority to support the customer service team.
Exercise

Paste something real

Two steps. First, paste something you've actually written. Then write two different instructions for the same material — and notice how different the outputs could be.

↩ From Lesson 11

Your brief from Lesson 11 is already a great example of this pattern — it's a context block ready to use. Now you're doing the same thing with your own content. Same formula, different material.

1
Paste your material
An email, a message, a document, meeting notes, a paragraph you've written — anything real. The longer the better, but even a few sentences works.
2
Write two different instructions
Same material, two completely different outputs. That's the point — you're showing yourself how much the instruction changes what comes back.
Knowledge check

Three quick questions

1. What's the key difference between asking the AI to generate vs asking it to process?
2. Which of these is a well-formed paste-and-instruct prompt?
3. You paste the same email into two prompts — one asks for a summary, the other asks for a rewrite. What's true?
+25
XP Earned
✦ 0 XP total
Next: Lesson 13 — The system prompt (a sneak peek) You've been building context manually. Next you'll learn that developers can pre-load it invisibly — and why that changes how you think about every AI tool you use.

You're processing, not just prompting 🔄

Paste-and-instruct is one of the most practical skills in this course. You'll use it constantly.