AI Intermediate

This week moves from individual prompts to systems. You will build a library of reusable templates and learn to chain prompts for complex tasks that a single prompt cannot handle.

The 80/20 rule of effective prompting

The fundamental insight: 80% of effective prompting is context and setup. Only 20% is the actual question.

Most founders jump straight to questions without giving AI the necessary context. They ask “Write me an email to a prospect” and wonder why the output is generic.

The fix: Build reusable context blocks that can be pasted before any question. This creates a library of Company Context Documents that dramatically improve every AI interaction.

Building your company context document

Create a single document containing:

Company: Name and stage (seed, Series A, etc.)

What we do: One sentence description of your business

Target customers: Clear definition of who you serve

Value proposition: Why customers should choose you

Competitors: Top three direct competitors

Traction: Current numbers (customers, revenue, growth metrics)

Key metrics: What you are focused on measuring

Voice: Brand personality and tone guidelines

This document gets pasted at the start of every prompt. AI immediately understands your context instead of starting from zero.

Practical prompt template categories

Communication prompts

Prospect research: Before every sales outreach, use AI to find pain points, how your product solves them, personal hooks, decision makers, and opening lines. Transforms 30 minutes of research into 5 minutes.

Cold emails: Personalized outreach using company context and prospect research combined.

Follow-up emails: Strategic sequences that add value without being pushy.

Social media posts: Voice-consistent content creation across platforms.

Business prompts

Decision framework: Structured thinking through hiring, pricing, pivots, partnerships, or fundraising decisions. Get pros, cons, hidden risks, alternatives, and reversibility analysis.

Customer discovery: Interview question generation and feedback analysis.

Competitor analysis: Market positioning, gaps, differentiation strategy.

Pricing research: Competitive analysis, margin health, price optimization.

Content creation prompts

Blog posts, LinkedIn articles, newsletters, product descriptions, testimonial posts - all following your voice profile.

Operations prompts

SOP creation: Step-by-step instructions with quality checks and troubleshooting guides for any repeatable process.

Prompt chaining for complex tasks

Single prompts work for simple questions. Complex tasks need chains - a series of prompts where each builds on the previous output.

The concept

Instead of asking AI one massive question, break it into linked steps:

1. Each step uses the output from the previous step as input
2. You review and adjust between steps
3. The final output is deeper, more connected, and more controlled

Example: Customer discovery analysis

Step 1 - Find pain patterns: “Here are transcripts from five customer calls. What are the common pain points mentioned?”

Step 2 - Rank by frequency: “Looking at these patterns, what are the MOST frequently mentioned pains? Rank them.”

Step 3 - Generate insights: “Based on this ranking, summarize key learnings and generate 10 refined questions for next interviews.”

Example: Go-to-market strategy

Step 1 - Analyze positioning: “Here are my competitors. Analyze their positioning, language, and messaging.”

Step 2 - Find gaps: “Based on this analysis, what customer needs are NOT addressed by any competitor?”

Step 3 - Define differentiation: “How can we own one of these gaps? Be specific.”

Step 4 - Define ideal customer: “Given this positioning, create an ideal customer profile.”

Step 5 - Find channels: “For this ICP, what are proven channels we could use? What would be differentiated?”

Step 6 - Design messaging: “Create copy and asset concepts for each recommended channel.”

Why chaining works

Avoids incorrect assumptions: Each step validates and refines the last before moving forward.

Gives you control: You can review, adjust, or redirect between any steps.

Produces deeper output: Building incrementally creates more connected results than a single massive prompt.

More effective than generic approaches: Each chain is tailored to your specific situation.

MCP: Connecting AI to your apps

MCP (Model Context Protocol) is a standard that allows AI to connect directly to your apps without copy-pasting.

Without MCP (manual workflow)

1. Copy email from Gmail
2. Paste into ChatGPT
3. Ask for analysis
4. Copy response back

Slow, manual, repetitive.

With MCP (connected workflow)

1. Ask: “Summarize unread emails from prospects”
2. AI connects to Gmail directly
3. AI reads and analyzes
4. AI returns summary

Automatic, fast, seamless.

MCP use cases

Gmail integration:

• Summarize unread emails from prospects
• Draft replies to customer inquiries
• Find all emails about a specific topic and summarize
• What did I promise to follow up on this week?

Google Drive integration:

• Summarize all docs in “Customer Interviews” folder
• Find inconsistencies across sales materials
• Which prompt templates am I missing?
• What decisions did customers make last month?

Other integrations:

• Slack: What did my team discuss about the product last week?
• Notion: From all my meeting notes, what is most critical this week?
• Calendar: Full background research on everyone I am meeting tomorrow
• Database: Which signups last month were active two weeks later?

The advantage

MCP lets you access data that is hard to upload or reference manually. Instead of finding, copying, and pasting context, AI retrieves what it needs directly.

AI image generation for founders

Beyond text, AI can generate branded visual content:

The formula

Subject + Style + Setting + Lighting + Mood + Technical specs + Brand

Why it matters for founders

• Generate images in seconds vs waiting days for a designer
• Unlimited iterations until it is right
• Perfect brand match when you include brand guidelines
• Significantly cheaper than custom design
• Ideal for social media, website heroes, blog headers, product mockups, pitch decks

Example prompts by use case

Hero banner: “Elegant fine dining table with burgundy candles, golden hour lighting, luxurious interior, wide angle, cinematic, 16:9”

Product shot: “Single velvet emerald candle on white pedestal, studio lighting, soft shadow, white background, commercial photography”

Lifestyle: “Cozy dinner party, friends at candlelit table, warm intimate atmosphere, bokeh lights, editorial lifestyle, warm tones”

Social media: “Flat lay candle making supplies, wax, wicks, pigments, marble surface, overhead, natural daylight, 1:1 square”

Pro tip

Create a style guide prompt with your brand colors, lighting preferences, and aesthetic guidelines. Paste it before every image request for consistent brand imagery across all generated visuals.

Building your prompt library

Organize your library into two main sections:

Personal prompts

• Voice profile (from Week 3)
• Background and bio
• Communication templates (emails, social posts)
• Content creation templates (blog, newsletter)

Business prompts

• Company context document
• Prospect research template
• Cold email sequences
• Follow-up templates
• Customer discovery questions and analysis
• Marketing templates (product descriptions, testimonial posts)
• Operations templates (SOP creation)
• Strategy templates (competitor analysis, pricing research)

Key takeaway

Stop writing prompts from scratch every time. Build a library of context and templates that compound in value. When 80% of the work is ready to paste, every AI interaction becomes dramatically more effective.