Framework
Follow along with the written framework below as you watch the curated video
What You'll Learn
- The anatomy of an effective prompt
- Why vague prompts = vague outputs
- The "Founder Prompt Formula"
- How to use examples to guide Claude
- When to iterate vs. start over
Why This Matters (2 min read)
The Scenario:
Founder: "Claude, help me with my product spec."
Claude: "I'd be happy to help with your product spec! Could you provide more details about:
- What product are you building?
- Who is it for?
- What problem does it solve?
- What features do you want to include?
- What's your timeline and budget?"
Founder: [Frustrated] "Why can't Claude just... know what I need?"
The Reality:
Claude is VERY capable — but it's not psychic.
Think of it like hiring a consultant:
❌ Bad Brief: "Help me with marketing."
- Consultant: "Uh... what kind of marketing? For what product? To whom?"
✅ Good Brief: "I need landing page copy for a B2B SaaS tool (team collaboration) targeting remote teams of 10-50 people. Goal: Get email signups for beta. Tone: Professional but friendly."
- Consultant: "Got it. Here's a draft..."
Same with Claude.
Vague prompt = Claude asks 10 clarifying questions. Specific prompt = Claude delivers immediately.
Watch This First: Prompt Engineering Basics (20 min)
Video: "Prompt Engineering for Beginners"
Source: DeepLearning.AI (Andrew Ng) Duration: 20 minutes (recommended excerpt) Why This Video: Clear, structured, covers fundamentals
[EMBED VIDEO HERE]
Key Timestamps:
- [02:30] What makes a good prompt
- [05:45] The power of examples (few-shot prompting)
- [08:15] Being specific vs. being verbose
- [11:20] Iterative refinement
- [14:40] Common mistakes
- [18:00] When to use system prompts
Watch the video, then come back for founder-specific applications below.
What You Just Learned (Summary)
From the video, you learned:
✅ Be specific — Tell Claude exactly what you want ✅ Provide context — Who, what, why, for whom ✅ Use examples — Show Claude what "good" looks like ✅ Set constraints — Length, tone, format, what NOT to do ✅ Iterate — Refine output with follow-ups, don't start over
Now let's apply these to FOUNDER work (not dev examples).
The Founder Prompt Formula
Use this structure for every prompt:
[CONTEXT] + [TASK] + [CONSTRAINTS] + [FORMAT] = Good Prompt
1. CONTEXT (Who, What, Why)
Tell Claude:
- Who you are (role: founder, PM, etc.)
- What you're building (product, industry, stage)
- Why you're asking (goal: validate, launch, scale)
Example:
textI'm a non-technical founder building a B2B SaaS tool for remote teams (10-50 people). Product: Team collaboration platform (like Slack + Asana). Stage: Pre-launch, validating product-market fit.
2. TASK (What You Want)
Be explicit about what you need Claude to do.
❌ Bad (Vague):
Help me with my product.
✅ Good (Specific):
Write a product spec for our "Task Automation" feature.
3. CONSTRAINTS (Rules & Limits)
Tell Claude:
- Length (1 page? 500 words?)
- Tone (professional? casual? technical?)
- What to include (must-haves)
- What to exclude (avoid jargon, no code, etc.)
Example:
textConstraints: - Keep it under 2 pages (founders don't read 10-page specs) - Use plain English (no technical jargon) - Focus on user benefits (not technical implementation) - Include: Problem, solution, key features, success metrics
4. FORMAT (How Output Should Look)
Specify structure:
- Bullet points vs. paragraphs
- Sections/headings
- Examples or tables
Example:
textFormat: ## Problem Statement [Describe the problem in 2-3 sentences] ## Proposed Solution [How this feature solves it] ## Key Features - [Feature 1] - [Feature 2] - [Feature 3] ## Success Metrics [How we'll measure success]
Putting It Together: Before vs. After
Example 1: Product Spec Generation
❌ BAD PROMPT (Vague)
Help me write a product spec for a new feature.
Why it's bad:
- What feature?
- What product?
- Who's the audience?
- What format?
Claude's response:
textI'd be happy to help! Could you tell me: - What product are you building? - What's the feature? - Who will use it? - What problem does it solve?
You wasted a turn. Now you need to answer 4 questions.
✅ GOOD PROMPT (Specific)
text[CONTEXT] I'm a non-technical founder building a B2B SaaS tool for remote teams (10-50 people). Product: Team collaboration platform (Slack + Asana hybrid). Stage: Pre-launch MVP. [TASK] Write a product spec for our "Task Automation" feature. This feature lets users create automated workflows (e.g., "When task is marked Done, notify team in Slack"). [CONSTRAINTS] - Keep it under 2 pages - Use plain English (no technical jargon) - Focus on user benefits (not technical implementation) - Target audience: Non-technical team leads [FORMAT] Use this structure: ## Problem Statement [2-3 sentences] ## Proposed Solution [How this feature works] ## Key Features - [Feature 1] - [Feature 2] ## Success Metrics [How we measure success] ## User Stories - As a [role], I want to [action] so that [benefit]
Claude's response:
text## Problem Statement Remote teams waste 5-10 hours per week on repetitive coordination tasks: manually notifying teammates when tasks are done, updating multiple tools when status changes, and chasing down approvals. This friction slows down project velocity and frustrates team leads who want to focus on strategy, not busywork. ## Proposed Solution Task Automation lets users create "if-this-then-that" workflows without code... [FULL SPEC DELIVERED IMMEDIATELY]
Why it's good:
- Claude has ALL the context upfront
- No back-and-forth
- Output matches your format exactly
- Tone is non-technical (as requested)
Example 2: Customer Research Synthesis
❌ BAD PROMPT
Analyze these customer interviews.
Why it's bad:
- What are you looking for?
- How many interviews?
- What format for output?
✅ GOOD PROMPT
text[CONTEXT] I conducted 10 customer interviews (remote team leads, 10-50 person companies). Goal: Understand their biggest pain points with current collaboration tools. [TASK] Analyze these 10 interview transcripts (in research/interviews/) and identify: 1. Top 5 pain points mentioned most often 2. Patterns in how they describe the problem 3. What solutions they've tried (and why they failed) [CONSTRAINTS] - Focus on pain points (not feature requests) - Rank by frequency (how many interviewees mentioned it) - Include direct quotes to support each finding [FORMAT] ## Pain Point 1: [Name] - Mentioned by: X/10 interviewees - Description: [2-3 sentences] - Quote: "[Direct quote from interview]" - Why current solutions fail: [1-2 sentences] [Repeat for Pain Points 2-5] ## Summary [3-4 sentences: What this tells us about product direction]
Claude's response:
text## Pain Point 1: Context Switching Between Tools - Mentioned by: 9/10 interviewees - Description: Teams use 5-8 different tools (Slack, Asana, Google Docs, etc.) and waste time switching between them, losing context, and duplicating work... - Quote: "I spend more time updating tools than doing actual work." — Interview #3 - Why current solutions fail: Integrations exist but require technical setup... [FULL ANALYSIS DELIVERED]
Example 3: Landing Page Copy
❌ BAD PROMPT
Write landing page copy for my product.
✅ GOOD PROMPT
text[CONTEXT] Product: TeamFlow (B2B SaaS, team collaboration + automation) Target: Remote teams, 10-50 people, non-technical team leads Stage: Beta launch (goal: 100 email signups) [TASK] Write landing page copy (hero section + 3 benefit sections + CTA). [CONSTRAINTS] - Hero: 1 headline + 1 subheadline (benefit-focused, not feature-focused) - Benefits: Focus on time saved, not technical features - Tone: Professional but approachable (like Slack, not like SAP) - Length: ~200 words total [FORMAT] ## Hero **Headline:** [6-8 words, benefit-driven] **Subheadline:** [15-20 words, specific outcome] ## Benefit 1: [Title] [2-3 sentences] ## Benefit 2: [Title] [2-3 sentences] ## Benefit 3: [Title] [2-3 sentences] ## CTA [Button text + supporting text] [EXAMPLE OF GOOD COPY] Headline: "Stop Managing Tools. Start Managing Teams." Subheadline: "TeamFlow automates the busywork so you can focus on what matters."
Claude's response:
text## Hero **Headline:** Stop Managing Tools. Start Managing Teams. **Subheadline:** TeamFlow automates coordination tasks and saves your team 10+ hours per week—no technical setup required. ## Benefit 1: Automate Repetitive Work Your team spends hours updating Slack, Asana, and email when tasks change. TeamFlow connects your tools and handles updates automatically... [FULL LANDING PAGE DELIVERED]
The Power of Examples (Show, Don't Just Tell)
Claude learns FAST from examples.
Technique: "One-Shot Prompting"
Show Claude ONE example of what you want:
textWrite 5 social media posts for our product launch. Example of a good post: "Tired of tool overload? TeamFlow connects your team's favorite tools so you can stop switching tabs and start shipping. Try it free → [link]" Tone: Casual, benefit-focused, under 200 characters. Now write 5 more like this.
Claude will mimic:
- The tone (casual)
- The structure (problem → solution → CTA)
- The length (~200 chars)
When to Iterate vs. Start Over
Iterate When (Refine Output):
✅ Output is 80% right — Just needs tweaks ✅ Tone is off — "Make this more casual" ✅ Length is wrong — "Shorten to 500 words" ✅ Missing one section — "Add a pricing comparison table"
How to iterate:
textClaude, this is great! Two changes: 1. Make the tone more casual (less corporate) 2. Add a section on pricing vs. competitors
Start Over When (New Prompt):
❌ Output is completely wrong — Claude misunderstood the task ❌ You forgot key context — Didn't mention target audience ❌ Format is unusable — Claude wrote paragraphs, you need bullets
How to start over:
textLet's start fresh. Ignore previous output. [NEW PROMPT WITH FULL CONTEXT]
Common Prompting Mistakes (Founders Make)
Mistake 1: Too Vague
❌ "Help me with marketing" ✅ "Write landing page hero copy for a B2B SaaS tool targeting remote teams"
Mistake 2: Too Much Context (Overload)
❌ "Here's 10 pages of background... [giant essay]... now write a product spec" ✅ "Here's the key context: [3 bullet points]. Now write a product spec."
Mistake 3: No Format Specified
❌ "Analyze these interviews" ✅ "Analyze these interviews and output in this format: [show structure]"
Mistake 4: Assuming Claude Knows Your Product
❌ "Generate a spec for the automation feature" ✅ "I'm building [product]. Generate a spec for [feature that does X]."
Mistake 5: Not Using Examples
❌ "Write in a casual tone" ✅ "Write in a casual tone like this: [show example]"
Try This Now (Exercise — 10 Minutes)
Step 1: Write a BAD Prompt (2 min)
Pick a task (product spec, customer analysis, landing page). Write a deliberately vague prompt (like the "bad" examples above).
Step 2: Use the Formula (5 min)
Rewrite using the Founder Prompt Formula:
- [CONTEXT] — Who you are, what you're building
- [TASK] — What you need
- [CONSTRAINTS] — Rules, limits, tone
- [FORMAT] — Structure of output
Step 3: Compare Outputs (3 min)
- Ask Claude your BAD prompt
- See how Claude responds (probably asks clarifying questions)
- Ask Claude your GOOD prompt
- See the difference (immediate, high-quality output)
What You Learned:
- Specificity = speed
- Context upfront = no back-and-forth
- Format guidance = usable output immediately
Downloads
📥 Founder Prompt Formula Template — Fill-in-the-blank structure 📥 10 Prompt Examples (Product, Research, Marketing) — Copy-paste ready 📥 Before/After Prompt Gallery (PDF) — Visual examples 📥 Prompting Checklist — Did you include context/task/constraints/format?
What's Next
You now know HOW to prompt Claude effectively. But there's a SECRET WEAPON that makes prompting even easier: CLAUDE.md files.
In the next lesson, we'll cover:
- What CLAUDE.md is (and why every project needs one)
- How to write a CLAUDE.md for YOUR work
- Industry-specific templates (SaaS, agency, e-commerce)
- How CLAUDE.md saves you from repeating context
→ Next Lesson: CLAUDE.md Setup & Project Context
Key Takeaways
✅ Use the Founder Prompt Formula: [Context] + [Task] + [Constraints] + [Format] ✅ Be specific, not vague: "Write landing page hero copy" > "Help with marketing" ✅ Show examples: Claude learns from what you show, not just what you tell ✅ Iterate on good outputs: Refine vs. starting over ✅ Front-load context: Save time by giving Claude everything upfront
Sources & Attribution:
This lesson curates content from:
- DeepLearning.AI: Prompt Engineering for Developers (Video) — Embedded tutorial
- Anthropic: Prompt Engineering Overview — Core principles
- Prompt Builder: Claude Best Practices 2026 — Modern techniques
Additional insights and founder-specific examples by NerdSmith.
Version: 1.0 (2026-02-08) Module: 0 — Claude Code Bootcamp Lesson: 3 of 5
Downloadable Resources
Click any resource to download. All downloads are free for enrolled students.