Better AI Prompts · Level 1: Prompting Basics · Lesson 3
Add the right context and constraints
AI models respond to context and constraints. Without constraints, an assistant might write in a generic tone or invent missing facts. In this lesson, you will learn how to set boundaries that keep AI answers relevant, accurate, and concise.
By the end of this lesson, you can:
- Distinguish helpful context from unnecessary information overload.
- Apply tone controls and audience scoping to match your communication style.
- Ask the assistant to avoid unsupported details, and check whether it followed that instruction.
Why context transforms AI responses
AI models do not possess background knowledge about your specific project, your team's culture, or your client's preferences unless you provide it. When you omit context, the AI falls back on generic averages derived from its training data.
Low context
Generic baseline
Write an apology email for a late shipment.
Produces a generic, overly formal template that sounds like a standard corporate auto-responder.
Scoped context & constraints
Tailored result
Write an apology email for an artisan coffee roastery. A courier truck broke down, delaying 40 orders by 24 hours. Keep it warm, informal, and offer a free bag of coffee on their next order. Do NOT use legal language or formal jargon.
Gives the assistant a tone and specific facts to work with. Check the wording and confirm that any offer is authorised before sending.
The power of negative constraints
You can describe both what you want and what the assistant should leave out. These instructions can help, but they do not prevent mistakes. Check the response against your source:
- Preventing Hype: "Do not use buzzwords like revolutionary, seamless, or game-changing."
- Discouraging invented details: "Do not invent facts, dates, or prices not provided in the source text."
- Controlling Length: "Do not write more than 150 words."
- Eliminating Filler: "Do not include introductory pleasantries (e.g. 'Sure, I can help with that!')."
Avoiding context overload
While context is crucial, dumping thousands of words of irrelevant background information can confuse the model or cause it to overlook your core instruction. Apply the Need-to-Know Rule:
Include only the background details that directly influence the required output. If a detail does not change what the AI should write, edit it out before sending your prompt.
Illustrative example
Prompt Makeover
A product manager wants to announce a new software update to existing beta testers. See how adding negative constraints and audience context prevents generic hype.
Original prompt
Write an email telling users about our new dark mode feature.What is missing
- Target audience: who is receiving this email?
- Tone: what voice should be used?
- Negative constraints: what buzzwords or claims should be avoided?
- Call to action: what specific step should users take next?
Improved prompt
Write a short announcement email for our technical beta testers about the release of Dark Mode v2.
Use a direct, helpful tone. Keep the message under 120 words.
CONSTRAINTS:
- Do NOT use marketing hype, clichés, or words like 'game-changing', 'revolutionary', or 'supercharge'.
- Do NOT invent unstated features; mention only custom themes and reduced eye strain in low-light environments.
- End with a single CTA link asking users to enable dark mode under Account Settings.Why it is better
- It scopes the audience: technical beta testers.
- It sets explicit negative constraints eliminating marketing clichés.
- It bounds the facts so the AI does not invent unsupported sub-features.
- It specifies an exact word limit (<120 words) and a single call to action.
What still needs checking
- Confirm that Dark Mode v2 is actually located under Account Settings.
- Ensure the tone reflects your company's actual brand voice.
Practice
Identify the prompt with effective negative constraints
Read both instructions for drafting a client proposal summary. Select the option that uses negative constraints to protect fact accuracy.
Illustrative example
A freelancer is summarizing a client scope agreement. They need a 1-page overview without added promises or estimated completion dates.
Check the model answer
Option B uses explicit negative constraints.
Option B explicitly tells the AI what NOT to do ('Do NOT state an estimated delivery date unless explicitly in the text'). Option A encourages the AI to be 'optimistic', which risks inventing imaginary timelines or over-promising.
Knowledge check
Check the answer
Option B is the correct answer.
Negative constraints (e.g. 'Do NOT use jargon', 'Do NOT assume unstated dates') describe what to leave out. The assistant can still make mistakes, so compare its answer with your source.
Keep this
The Context & Constraints Formula
Combine positive context with negative boundaries to keep responses on target:
TASK: [What to produce]
AUDIENCE: [Who will read/use this]
TONE: [Direct / Professional / Friendly / Concise]
MUST INCLUDE: [Essential facts, links, or details]
DO NOT INCLUDE: [Prohibited buzzwords, assumed dates, or unverified claims]Negative constraints are especially powerful when writing customer-facing copy, legal summaries, or technical guides where accuracy is critical.
Optional browser progress
Your progress
0 of 5 available lessons completed
Progress stores only the course identifier and completed lesson identifiers in this browser. It does not save your answers, prompt text or identity, and it does not sync to another device. Clearing browser data removes it.