Output Formats & Constraints — Foundations Lesson 10
Scenario
You ask ChatGPT: "Give me a summary." It gives you 5 rambling paragraphs when you needed 3 bullets. You ask: "Create a timeline." It lists events chronologically but includes irrelevant details. You realize: I haven't been telling AI how to format what it gives me. Once you specify format — bullets, structured tables, word count, tone — AI produces exactly what you need. Suddenly, outputs are usable without refinement.
---
Learning Objectives
By the end of this lesson, you will be able to:
1. Specify output formats effectively (bullets, tables, paragraphs, etc.)
2. Use constraints to control response length and detail
3. Structure outputs for your specific use (email, report, presentation)
4. Define success criteria through format specification
5. Get "right-sized" outputs without excessive refinement
---
Think Before You Format
Before you write a prompt, ask:
---
Output Formats: Choosing the Right One
Format 1: Bullet Points
When to use: Summarizing, listing options, capturing key points, quick reference
Structure:
`
[Topic]
`
Example prompt:
`
Summarize the key risks of expanding into the Asian market
in bullet points. No more than 5 bullets.
`
Why it works: Bullets are fast to read, easy to skim, good for executive presentations
When NOT to use: When you need full context or detailed reasoning (use paragraphs instead)
---
Format 2: Structured Paragraphs
When to use: Full explanation, narrative flow, complete reasoning, writing that needs flow
Structure:
`
Paragraph 1: [Topic introduction]
Paragraph 2: [First idea / context]
Paragraph 3: [Second idea / development]
Paragraph 4: [Conclusion / recommendation]
`
Example prompt:
`
Write a 4-paragraph email explaining why we're switching vendors.
Each paragraph should cover: why change, what's broken now,
how new vendor fixes it, next steps.
`
Why it works: Full context and reasoning. Reads naturally. Good for formal communications.
When NOT to use: When you need quick reference (use bullets instead)
---
Format 3: Tables
When to use: Comparing options, organizing data, side-by-side analysis
Structure:
`
| Column 1 | Column 2 | Column 3 |
|----------|----------|----------|
| Data | Data | Data |
`
Example prompt:
`
Compare three project management tools (Asana, Monday, Jira)
in a table with rows: pricing, learning curve, best for, drawbacks.
`
Why it works: Easy comparison. Organized. Scannable for decision-making.
When NOT to use: When you need narrative explanation (tables don't show why, just what)
---
Format 4: Numbered Steps/Procedure
When to use: Instructions, processes, sequences, workflow
Structure:
`
1. [Step 1]
2. [Step 2]
3. [Step 3]
`
Example prompt:
`
Give me 6 steps to onboard a new customer, starting with
initial contact and ending with first purchase.
`
Why it works: Clear sequence. Easy to follow. Good for training.
When NOT to use: When order doesn't matter (use bullets instead) or when each step has complex explanation (use paragraphs)
---
Format 5: Outline
When to use: Planning complex documents, organizing information, showing structure
Structure:
`
I. Main Topic
A. Subtopic 1
1. Detail
2. Detail
B. Subtopic 2
II. Main Topic 2
`
Example prompt:
`
Create an outline for a presentation on AI in business,
with 3 main topics, 2-3 subtopics per main topic, and
2-3 supporting points per subtopic.
`
Why it works: Shows hierarchy and structure. Good for planning before writing full content.
When NOT to use: When you need final output (outlines are planning tools, not finished work)
---
Format 6: Structured Data (JSON, XML, CSV)
When to use: Data that needs to be processed or imported
Structure:
`json
{
"key": "value",
"nested": {
"item": "data"
}
}
`
Example prompt:
`
List 5 customer segments in JSON format with fields:
segment_name, size, avg_spending, primary_pain_point.
`
Why it works: Structured data can be imported into tools, databases, or systems
When NOT to use: When humans are reading it (use tables or bullets instead)
---
Constraint Types: Controlling Output Characteristics
Constraint 1: Length
Specify:
Example prompt:
`
Summarize the Q3 financial report in under 300 words.
Include revenue, profit, and three key metrics.
`
Why it works: Prevents rambling. Forces conciseness. Makes output scannable.
---
Constraint 2: Detail Level
Specify:
Example prompt:
`
Explain how our recommendation engine works.
Executive summary level (can be explained in 1 paragraph
to someone unfamiliar with machine learning).
`
Why it works: AI calibrates depth to your needs. Prevents both oversimplification and over-explanation.
---
Constraint 3: Tone/Style
Specify:
Example prompt:
`
Draft an email to a frustrated customer.
Tone: warm, apologetic, but not overly formal.
Acknowledge their frustration without making excuses.
`
Why it works: AI adjusts language choice to match tone. Same content, different feel.
---
Constraint 4: Audience
Specify:
Example prompt:
`
Explain why we're switching to cloud infrastructure.
Audience: non-technical executives who care about cost and security.
Avoid technical jargon. Focus on business impact.
`
Why it works: AI tailors language and emphasis to audience. Executives get business impact; engineers get technical details.
---
Constraint 5: Included/Excluded Content
Specify:
Example prompt:
`
Propose a new customer onboarding workflow.
Must include: email automation, welcome call, resource guide.
Must NOT include: any changes that require IT involvement.
Should focus on: reducing time-to-first-value.
`
Why it works: Prevents AI from going off-topic. Ensures key points are covered.
---
Building Effective Output Specifications
The Complete Format Specification
`
Format: [Format type - bullets, paragraphs, table, steps, etc.]
Length: [Specific word count, number of items, or time to read]
Detail: [Level of detail - executive, middle-ground, detailed]
Tone: [Communication tone - formal, warm, technical, casual, etc.]
Audience: [Who reads this - executives, team, clients, etc.]
Must include: [Essential elements]
Must NOT include: [What to avoid]
`
Example:
`
Format: Bullet points
Length: 5 key points maximum, under 150 words total
Detail: Executive summary (no technical jargon)
Tone: Professional but warm
Audience: Our board of directors (non-technical)
Must include: ROI impact, timeline, resource requirements
Must NOT include: detailed technical architecture
`
---
Common Format Mistakes
Mistake 1: No Format Specified
AI gives rambling 3-page response.
AI gives concise, scannable list.
---
Mistake 2: Vague Length Constraints
AI doesn't know if you want 1 paragraph or 10.
AI knows exactly how much to write.
---
Mistake 3: Not Specifying Detail Level
AI might assume you know nothing, or everything.
AI calibrates correctly.
---
Mistake 4: Competing Constraints
Impossible. Comprehensive = long.
Clear trade-off: depth for conciseness.
---
Mistake 5: Not Stating What Should Be Included
AI might create generic values that don't match your actual culture.
AI knows what's essential.
---
Mission: Format Specification Practice
The Task
1. Pick 3 different outputs you want AI to create (email, analysis, plan, summary, etc.)
2. For each output, write a complete format specification using the template
3. Test with an AI tool — how well did the format spec work?
4. Refine — what would you change about the spec?
Example outputs to consider:
---
Deliverables
Format Specification Workbook:
`
OUTPUT 1: [Name]
Format Specification:
Prompt You'll Use:
[Your complete prompt with specification]
Result:
[What AI created]
[Was the format correct?]
[What worked? What didn't?]
Refined Specification:
[What you'd change for next time]
[Repeat for outputs 2 and 3]
`
Document your learning. You're building format fluency.
---
AI Coach: Format Specifications Are Prompts
Here's what many people miss: Your format specification is your prompt.
A vague prompt + detailed format spec = Better output than a detailed prompt + no format spec.
Why? Because AI responds to specificity. When you say "2-3 bullets, under 100 words, formal tone," you're giving AI clear targets to hit.
Key insight: Spend less time on flowery prompt language, and more time on specific format requirements. AI respects structure.
---
Reflection: Your Preferred Output Formats
Different tasks call for different formats. What's your default?
Reflect:
Write 3-4 sentences about your format preferences.
---
Portfolio Check
Portfolio principle: Every piece in your portfolio should have intentional format.
This means:
Format choices are professional choices.
---
Bonus Challenge
Format audit: Take three pieces of writing you've done (emails, reports, etc.). For each, what format did you use? Could a different format make it better? This trains you to think about format intentionally.
---
Key Takeaways
Next: Lesson 11 explores Building AI-Assisted Workflows — combining multiple techniques into repeatable processes.