The Complete Copywriting Guide for AI Founders Selling to Enterprises

Why Copywriting Matters for AI Founders

Based on Harry Dry's Masterclass, Contextualized for Enterprise AI Sales

Copywriting is the number one skill in marketing. It's not about tricking people—it's the art of simple, clear communication that resonates instantly.

Consider this: Most AI platforms do roughly the same thing. They promise automation, efficiency, intelligence. Strip away the branding and positioning, and you've got similar capabilities. Yet some AI companies command 10x valuations and close enterprise deals in weeks while others languish in pilot purgatory for years.

The difference? Words. Positioning. How you make the buyer feel.

"We don't choose the whiskey, we choose the image." — David Ogilvy

Enterprise buyers aren't evaluating your model's parameters. They're evaluating whether you understand their world, whether you'll make them look smart to their board, whether betting on you feels safe. That's copywriting.

The Three Rules of Great Copy

Apply these three questions to every sentence you write. Three yeses means you're onto something. Three nos means you've probably written rubbish.

Rule 1: Can I Visualize It?

If readers can't see it in their mind, they won't remember it.

"If I can't see it, it's not there yet." — Lisa Cron

The Test: Read these phrases and notice which ones stick:

• Seamless integration
• AI-powered insights
• Your logistics manager finally sleeping through the night
• End-to-end automation
• The CFO showing your dashboard in the board meeting

You remember the logistics manager sleeping. You remember the CFO in the boardroom. The others are abstract wallpaper—the kind of language every AI company uses.

Enterprise AI Examples:

‍

Abstract (Forgettable)Concrete (Memorable)Seamless integrationPlugs into SAP in 3 hours, not 3 monthsAI-powered insightsFlags the shipment that'll miss its window before your dispatcher knowsStreamlined workflowsYour warehouse picker finds items in 30 seconds instead of 4 minutesEnhanced efficiencyOne demand planner doing the work of four, home by dinnerDigital transformationThe CEO demo-ing your product to the board himself

Concrete things can be dropped on your foot. Abstract things just fade away.

How to Go From Abstract to Concrete:

Draw a line down a sheet of paper. Write the abstract word at the top. Keep asking: What do I actually mean? Keep rewriting until you land on something you can see.

Example: "Accelerate order fulfillment" → What does that mean? → Faster picking → How much faster? → 60% faster → What does that look like? → "Your picker finds the item in 30 seconds instead of wandering the warehouse for 4 minutes"

Remember "Couch to 5K"—the most downloaded fitness program ever, created by a Boston TV producer, not Nike. It stuck because it was concrete. Your enterprise AI pitch needs the same clarity.

Rule 2: Can I Falsify It?

Write sentences that are true or false. When you put your head on the chopping block with a falsifiable claim, people sit up and pay attention.

Galileo was sentenced to house arrest for saying the earth spins around the sun. If he'd said "the earth has a harmonious connection with a celestial object," they'd have told him to go have a beer.

The "Don't Talk, Only Point" Test:

Imagine you're describing your AI platform to a skeptical CIO. First, try it subjectively:

• It's cutting-edge
• It's enterprise-ready
• It delivers incredible ROI
• It's the best solution on the market

Now point to things that are true or false:

• Three Fortune 500 manufacturers run it in production
• Average deployment takes 42 days
• Customers see 23% reduction in inventory carrying costs within 6 months
• Our last three enterprise deals closed in under 90 days

The second version has teeth. Every claim can be verified. It gets you off the adjective trail.

Enterprise AI Application — Selling to Manufacturing:

Instead of saying "Our AI dramatically improves supply chain efficiency," point to:

• The graph showing stockouts dropping from 12% to 2%
• The specific manufacturer that reduced expedited shipping costs by $4.2M
• The exact dollar amount saved on a $200M inventory position
• The demand planner who said "I actually trust my forecasts now"

Pointing forces you to find concrete, falsifiable evidence. And enterprise buyers live on evidence.

Rule 3: Can Nobody Else Say This?

"Never write an ad a competitor can sign." — Jim Durfee

This rule forces you to look deeper at what makes your product unique. If Palantir, C3.ai, and every other horizontal platform could say the same thing, you haven't found your positioning yet.

The New Balance Lesson:

"Worn by supermodels in London and dads in Ohio."

Can you visualize it? Yes—the supermodel, the dad at the barbecue. Is it falsifiable? Yes—supermodels do wear New Balance, dads in Ohio do too. Can anyone else say it? No—Prada doesn't have dads in Ohio, Reebok doesn't have supermodels.

Enterprise AI Application:

Bad: "The leading AI platform for supply chain"

Good: "The only AI platform built by people who've actually walked a warehouse floor at 3am"

Bad: "Enterprise-grade security and compliance"

Good: "Deployed inside three manufacturers whose compliance teams rejected everyone else"

Bad: "Faster time to value"

Good: "Live in production while your Palantir POC is still in procurement"

Your Positioning Test:

What can you say that the horizontal players cannot?

• "Built by supply chain people, for supply chain people—not Silicon Valley tourists"
• "We don't do healthcare. We don't do finance. We do manufacturing. That's it."
• "The only AI platform where every engineer has stood on a factory floor"

Find the claim that's true, visual, and exclusively yours.

The Copywriting Process

Great copy doesn't emerge fully formed. It's built through iteration and structured thinking.

The Starting Point: Current → Desired

"The current attitude of the consumer is the starting point and the desired attitude is the finish line. You can't start a race in the middle."

Before writing anything, define:

• Point A: What does your enterprise buyer currently believe/feel/do?
• Point B: What do you want them to believe/feel/do?

Enterprise AI Example:

Point A (Current): "AI is overhyped. We've seen three failed POCs. My team is skeptical. I'll look stupid if this doesn't work. The horizontal platforms don't understand our business."

Point B (Desired): "These people actually get manufacturing. This isn't a science project—it's a production system. My team will look like innovators. This is the safe bet, not the risky one."

Your job is to string the wire between these two poles.

The Three Pieces of Great Copy

Piece 1: Who Are You Talking To?

Snapchat spent $7 million on a Super Bowl ad saying "More Snapchat, less social media." The average Super Bowl viewer is 39—someone just figuring out Facebook. That's not your Snapchat audience.

The Coffee Shop Test: Two cafes in St. Ives. One had a huge menu with 17 croissants and all their prices. The other just said: "Coffee and pastry = £5." Everyone queued outside the second one.

Why? They understood their audience: tourists walking, not paying full attention. They needed something that registers instantly.

Enterprise AI Application:

Who are you actually talking to?

• The CIO who's been burned by three failed AI initiatives and needs air cover
• The VP of Operations who's drowning in inefficiency and needs results by Q3
• The CEO who promised the board "digital transformation" and needs to show progress
• The Plant Manager who just wants to hit production targets without working weekends

Each needs different copy. The CIO needs safety and credibility. The VP needs speed and proof. The CEO needs a story for the board. The plant manager needs "your life gets easier."

Don't write for "manufacturing executives." Write for the specific person in the specific moment with the specific pain.

Piece 2: Have Something to Say

You need to believe something. Why did you start this company? What's broken that you're fixing?

Zuora's investors didn't believe in the product—they believed in the idea that everything was going to be subscription.

Enterprise AI Application:

What do you believe that your competitors don't?

• "Horizontal AI platforms will never work for manufacturing. The domain expertise isn't optional—it's everything."
• "The problem isn't the AI. The problem is that tech companies don't understand how factories actually run."
• "Every failed AI project in manufacturing failed because it was built by people who've never been on a production line."

This belief becomes your positioning. It becomes the thing that makes you impossible to ignore.

Your belief might be: "Manufacturing deserves AI built by manufacturing people, not Silicon Valley generalists playing tourist in a vertical they don't understand."

That's something to say.

Piece 3: Say It Well

This is what most people think copywriting is—the rhythmic, visual, persuasive part. But it's actually just the final piece.

Hinge Example: The brief was "an app for people who want long-term relationships and are fed up of dating apps." Not catchy.

The result: "The dating app designed to be deleted."

Notice the alliteration (DDD: dating, designed, deleted) and the juxtaposition (who makes an app that wants to be deleted?). No other app can say this now.

Enterprise AI Application:

Brief: "An AI platform that helps manufacturers optimize their supply chain"

Boring version: "AI-powered supply chain optimization for modern manufacturing"

Better versions:

• "The AI platform that actually ships. In production at factories you've heard of."
• "Manufacturing AI that works. Not a POC. Not a pilot. Production."
• "Built by ops people. Bought by ops people. Running in production at manufacturers that said no to everyone else."

Find the rhythm. Find the juxtaposition. Find the line that only you can say.

The Iteration Process: A Real Example

Great copy takes 20+ rewrites. Here's how it actually works.

The Seed:

"The difference between 1% and 2% is not just 1%, it's 100%."

If you can increase landing page conversions from 1% to 2%, you've doubled growth. There's something here.

Enterprise AI Version — The Seed:

"The difference between 94% on-time delivery and 99% on-time delivery isn't 5 points. It's the difference between losing your biggest customer and locking them in for a decade."

Finding the Conflict:

Draw a line. Write opposing ideas:

• Left side: Hire more planners, work longer hours, pray forecasts improve
• Right side: Deploy AI that sees what humans miss

The Iterations:

1. "Want to improve your on-time delivery? You've got two choices: hire more bodies or deploy smarter systems."
2. Added structure: "The Old Way vs. The AI Way"
3. "The Old Way: Hire 30 more demand planners at $120K each. Hope they're good. Wait 18 months to see results. The AI Way: Deploy in 42 days. See results in Q1. Pay for outcomes."
4. Tested with prospects. "Pay for outcomes" landed. The specificity of "42 days" and "$120K" added credibility.
5. Final: "Your competitors are hiring 30 more planners and hoping it works. You could be live in 42 days with AI that pays for itself."

Key Lessons:

• Start with a seed of an idea
• Find visual inspiration for structure
• Get feedback on multiple versions
• Keep iterating until you can't remove anything
• Specificity (42 days, $120K, 30 planners) beats vague claims

Classic Ads & What They Teach Enterprise AI Founders

"A thousand songs in your pocket" — Apple iPod

PC ads were all features: 4GB storage, processor speed. Apple ignored all that.

"Pocket" is metonymy—substituting a visual word for the literal "media player." It doesn't literally make sense. But it's more visual, more surprising, more punchy.

Enterprise AI Lesson: Stop selling features. Sell the transformation.

• Not: "ML-powered predictive analytics for demand planning"
• But: "Know which SKUs will stockout before your planner opens the spreadsheet"

"They don't write songs about Volvos" — Corvette

Six words. What can Corvette say that no other car can? That Post Malone sings about Corvettes. That country singers love them. It's romantic. It's theirs.

Enterprise AI Lesson: Find what's romantically, uniquely yours.

• "Nobody writes case studies about horizontal platforms. They write them about manufacturers who transformed their operations."

"I've never read The Economist. Management trainee, aged 42."

No picture. No logo. Just text on red. The visual hierarchy: "I've never read The Economist." Wait, what? "Management trainee, aged 42." Now there's a story.

Enterprise AI Lesson: Implied consequences beat stated benefits.

• Not: "Our AI helps you make better decisions"
• But: "Still forecasting in spreadsheets? Your competitor isn't."

Writing Simply

Kaplan's Law of Words

"Any words that aren't working for you are working against you."

Harry's corollary: You aren't taking Kaplan's law seriously enough. You can always cross something out.

Enterprise AI Application:

Before: "Our comprehensive, AI-powered platform leverages cutting-edge machine learning to deliver actionable insights that drive meaningful business outcomes across the manufacturing value chain."

After: "AI that catches what your planners miss. Live in 42 days."

The strength of an idea is inversely proportional to its scope. "We do manufacturing AI" beats "We do enterprise AI across manufacturing, healthcare, and financial services."

The Burrito Test

A good paragraph is like a burrito. Throw it to someone—they should catch it without it falling apart.

Pull one sentence out. A good paragraph should break. If you can remove a sentence and it still works, that sentence shouldn't have been there.

The Two-Line Rule

Keep paragraphs to two lines maximum. Three reasons:

1. Short paragraphs are easy to swing between. Executives read on planes, between meetings. Help them swing.
2. If a paragraph runs three lines, you're probably not explaining yourself as well as you could. Cut something.
3. Good lines need room to breathe.

The Power of Facts

When in doubt, lead with a fact. Facts guarantee you're saying something. Most marketing says nothing—wallpaper, word-shaped air.

Facts Create Precision:

Vague: "We help manufacturers improve performance."

Precise: "Manufacturers using our platform see an average 34% reduction in stockouts within 6 months. The best performer went from 12% stockout rate to under 2%."

Facts Create Stories:

The fact: "The average enterprise AI POC takes 9 months. Our average deployment takes 42 days."

The story: "While your competitors are still in procurement with Palantir, you could be in production."

Facts vs. Commentary:

Bad: "Our customers love us and see great results."

Good: "Three of our last five customers expanded to enterprise-wide deployment within 12 months. Our NPS is 74. Our average contract value grew 280% from year one to year two."

Facts have weight. Adjectives don't.

Creating Structure

Your standards are your work. Divide ideas into clear categories to make them digestible.

Dividing Lines and Parallelism

Abstract: "Here's why you should buy our platform."

Structured: "There are three reasons manufacturers choose us over horizontal platforms: domain expertise, deployment speed, and outcome-based pricing. Let me walk you through each."

Now there's shape. Progress. Something to hold onto.

The "Actually Two Things" Framework

Take one thing, show it's actually two.

"You want AI for your manufacturing business" → "Actually, that's two things: you want AI that works in a POC, and you want AI that works in production. Most platforms can do the first. Almost none can do the second. Let me tell you why they're completely different problems."

Now you have structure. Now you have a story.

Enterprise AI Messaging: Putting It All Together

For the CIO (Safety & Credibility)

Current state: "I've been burned by AI vendors. My board wants innovation but I need air cover. I can't afford another failed POC."

Message: "Three Fortune 500 manufacturers run our platform in production. Not pilots. Production. We're the safe choice that looks like the innovative choice."

For the VP of Operations (Speed & Proof)

Current state: "I'm drowning. My team is burning out. I need results by Q3 or I'm going to lose people."

Message: "Live in 42 days. Your planners confident in their forecasts instead of firefighting all day. That's not a promise—that's what happened at [Customer X]."

For the CEO (Board Story)

Current state: "I promised digital transformation. The board is asking for progress. I need something I can show."

Message: "Imagine walking into your next board meeting with a live demo of AI predicting demand in real-time. Not a roadmap. Not a vendor presentation. Your data. Your system. Live."

For the Plant Manager (Life Gets Easier)

Current state: "I'm exhausted. I fight fires every day. I work weekends just to hit targets."

Message: "What if you actually took a weekend off? What if the AI flagged problems before they became emergencies? That's not science fiction—that's what plant managers at [Customer X] do now."

On AI Writing Tools and Taste

"Switching from pens to typewriters didn't make the work better. Switching from typewriters to laptops didn't make the work better. AI is a tool. It's only as good as your taste."

What AI Can't Do:

• Taste: Ogilvy pulled "At 60 mph, the loudest noise from the Rolls-Royce is the electric clock" from a motor magazine. Why that line? Taste. AI doesn't know what to pull out.
• Conviction: AI doesn't believe anything. It can't put its head on the chopping block with a falsifiable claim it stands behind.
• Experience: Bukowski was a postman for 20 years before writing Post Office. Kerouac lived on the road for seven years before writing On the Road in three weeks. Michael Lewis worked on Wall Street before writing The Big Short.

What This Means for You:

Your experience in your vertical—the deals you've closed, the pilots you've seen fail, the operators you've watched struggle—that's your unfair advantage. AI can help you write faster. It can't replace knowing what it feels like to sit across from a skeptical CIO who's been burned three times.

Use AI to draft. Use your taste to edit. Use your experience to know what's true.

Quick Reference: The Enterprise AI Copywriting Checklist

Before You Write

• Define Point A (current attitude) and Point B (desired attitude)
• Know exactly who you're talking to (CIO? VP? Plant manager?)
• Know why your product matters—what you believe that competitors don't

For Every Sentence, Ask:

1. Can I visualize it? If not, make it concrete. "Planner home by 5pm" beats "improved efficiency."
2. Can I falsify it? If not, point to something true or false. "42 days to deployment" beats "fast implementation."
3. Can nobody else say this? If not, find what's unique. "Built by ops people" beats "enterprise-grade AI."

The One Mississippi Test

Show your copy to a prospect. One Mississippi, two Mississippi. Do they get it? If it takes longer than two seconds, keep iterating.

When Editing

• Apply Kaplan's Law: Any word not working for you is working against you
• Apply the Burrito Test: Can you remove a sentence without breaking the paragraph?
• Keep paragraphs to two lines or less
• Lead with facts when possible

Final Check

• Does it feel specific to your buyer's world?
• Can you remove anything else?
• Would a competitor be able to sign this? (If yes, rewrite)
• Is there a concrete image the reader can see?

‍

‍