10 //

Workflows & Automation

Claude Code basics, brand voice files, batch processing. Tools to scale beyond manual prompting.

10 // WORKFLOWS & AUTOMATION

Once you've mastered prompting, the next edge is speed.

How fast can you go from brief to 50 script variants? If the answer is "hours," you're leaving money on the table.

Here's how to automate it.


Basic Claude Code Usage

Most affiliates use Claude through the web interface. Type a prompt, get a response, copy-paste it somewhere. That works for one-off scripts. It doesn't work when you need 50 variations across 4 formats by end of day.

Claude Code is the terminal-based version of Claude. It runs in your command line and can read files, write files, and process batches. If you're spending more than an hour a day prompting through the web interface, Claude Code will cut that in half.

What you need:

  • A terminal (Mac: Terminal app. Windows: Command Prompt or WSL.)
  • A Claude account with API access
  • Basic comfort with typing commands (you don't need to be a developer)

What it lets you do that the web interface doesn't:

  • Read your brand voice file and apply it automatically to every prompt
  • Process a batch of briefs into finished scripts without copy-pasting
  • Save and reuse prompt templates as files
  • Output results directly to files you can share with your team

Basic workflow:

  1. Write your prompt template in a text file (e.g., video-script-template.txt)
  2. Write your campaign brief in another file (e.g., glp1-brief.txt)
  3. Ask Claude Code to read both files and generate the output
  4. Claude Code writes the result to a new file (e.g., glp1-scripts-v1.txt)

No copy-pasting. No switching between tabs. No reformatting.


Storing Brand Voice Files

This is the biggest time-saver nobody uses.

Every time you prompt Claude through the web interface, you have to re-explain your brand voice, your compliance constraints, your audience, your style preferences. Every single time. That's 30% of your prompt just repeating context Claude already had in the last conversation.

With Claude Code, you store that context in a file once and reference it forever.

Example: brand-voice.md

# PingProfit Brand Voice

## Tone
- Direct, no-BS, results-first
- Confessional (admits mistakes, shares real numbers)
- Peer-to-peer (not teacher-to-student)
- Short sentences, one idea per paragraph

## Compliance Boundaries (Health Offers)
- No guaranteed outcomes
- No medical claims
- Personal experience framing is okay
- Mechanism explanation is okay
- Doctor attribution is okay
- Must include soft CTA, never hard sell

## Audience
- Media buyers spending $1k+/day
- 25-45, technically capable, not developers
- Active on Money Twitter and Telegram
- Care about ROI, not theory

## Authenticity Markers (Always Include)
- Hesitation words in openings
- Self-deprecation when sharing failures
- Specific numbers for every claim
- Conversational asides
- Casual grammar (gonna, wanna, kinda)

## Kill List (Never Use)
- "Game-changing"
- "Transform your life"
- "Real results"
- "Are you struggling with..."
- "Journey"
- Em dashes
- Perfect grammar
- Questions as hooks

Example: glp1-offer.md

# SlimRx GLP-1 Offer Details

## Product
- Telehealth GLP-1 (semaglutide) injections
- Weekly self-administered injection
- Includes physician consultation
- $300-400/month (don't mention in ads)

## Payout
- $150 CPA

## Target Audience
- 35-55 years old
- HHI $75k+
- Previous dieters (keto, Whole30, WW, IF)
- Multiple failed attempts
- Tired of yo-yo cycling
- Heard about Ozempic, may be skeptical

## Compliance
Cannot say:
- Guaranteed weight loss amounts
- "Lose X pounds in Y days"
- Medical claims (treats/cures anything)
- Before/after numbers as promises

Can say:
- Personal experience
- General mechanism (hunger hormones, GLP-1 receptor)
- "My doctor recommended..."
- Individual results with disclaimer energy
- "Not for everyone"

## Winning Angles (from testing)
- Biology reframe ("not willpower, hormones")
- Doctor authority ("my actual doctor, not Instagram")
- Specific failure timeline ("6 years, 4 diets, same 20 lbs")
- Ozempic awareness ("yeah, that Ozempic thing")

Now when you want to generate creative, your prompt is:

Read brand-voice.md and glp1-offer.md.

Generate 5 new 30-second video scripts using the 30-second template
format from Section 04. Each script should use a different hook
formula. Apply all brand voice rules and compliance constraints
automatically.

That's it. Three sentences. Claude Code reads both files, applies all the context, and generates five scripts that match your voice, your offer, and your compliance boundaries. What used to take 30 minutes of prompt-writing takes 10 seconds.


Batch Processing

When you need volume, batch processing is how you get there.

The Brief-to-Batch Workflow

Step 1: Create a briefs file with multiple campaigns or angles.

# briefs.md

## Brief 1: GLP-1 Video Scripts (Confessional Angle)
- Format: 30-second UGC video
- Hook type: Confessional
- Quantity: 5 variants

## Brief 2: GLP-1 Video Scripts (Doctor Authority Angle)
- Format: 30-second UGC video
- Hook type: Doctor authority
- Quantity: 5 variants

## Brief 3: GLP-1 Image Ad Copy (Biology Reframe)
- Format: Facebook primary text + headline + description
- Angle: Biology reframe
- Quantity: 5 complete ad sets

## Brief 4: GLP-1 Advertorial Openings
- Format: Headline + first 3 paragraphs
- Style: Personal story
- Quantity: 3 variants

Step 2: Point Claude Code at it.

Read brand-voice.md, glp1-offer.md, and briefs.md.

Process each brief using the appropriate template format. Apply all
brand voice and compliance rules. Output each brief's results under
a clear header. Save everything to output/glp1-batch-[today's date].md

Step 3: Review and deploy.

18 pieces of creative. One command. Five minutes.

What to Batch

| Use Case | Batch Size | Time (Manual) | Time (Claude Code) | |----------|-----------|---------------|-------------------| | Hook testing | 10 hooks | 45 min | 2 min | | Full script variants | 5 scripts | 90 min | 5 min | | Image ad copy sets | 10 sets | 60 min | 3 min | | Headline A/B variants | 10 headlines | 30 min | 1 min | | Advertorial openings | 3 versions | 60 min | 3 min | | Cross-format expansion | 10+ pieces | 2+ hours | 5 min |

The math is simple. If you're generating creative daily (and you should be), Claude Code pays for itself in the first week.


Scaling Beyond Manual Prompting

The workflow we've described so far is the foundation. Here's where it goes from there.

Level 1: Manual prompting (web interface) You type prompts, get responses, copy-paste. This is where most affiliates are. It works but doesn't scale.

Level 2: Template-based prompting (Claude Code) You store brand voice, offer details, and templates as files. You reference them in short prompts. 5x faster than manual. This is where you should be after reading this guide.

Level 3: Batch automation You create brief files, process them in one command, output structured results. 10x faster than manual. This is what separates the affiliates doing $1K/day creative testing from the ones doing $100/day.

Getting from Level 1 to Level 2 is the biggest jump. It's also the easiest. You just need to write your brand voice file and your offer file once. After that, every prompt you write is shorter, faster, and more consistent.

Getting from Level 2 to Level 3 takes a bit more setup but the payoff is massive. Once your batch workflow is dialed, you can generate a full week's worth of creative variants in a single morning session.


Speed is the ultimate edge in performance marketing. The faster you can generate, test, and iterate, the faster you find winners. And the faster you find winners, the more money you make.

Set up your brand voice files today. Use Claude Code for your next creative batch. You'll never go back to the web interface for production work.