Patreon Premium Cheatsheet

5 Moves That Turn
Claude Into a Workforce

The 5 exact prompts I used to replace a $30,000 consulting team with one Claude session. Real customer project. Copy-paste ready.

By Rick Pina
Patreon.com/RickPina
5
Moves
8
Agents
1hr
Session
$30K
Replaced
Inside This Guide

Contents

00 Introduction: Read This First P. 03 01 Onboard Before You Ask P. 04 02 Promote It from Generalist to Specialist P. 05 03 Build a TEAM That Talks to Each Other P. 06 04 Save the Institutional Knowledge P. 07 05 Branded, Ready-to-Ship Deliverables P. 08 06 The Workflow Together P. 09 07 Quick Checklist P. 10
Introduction

Most People Use AI Like a Search Engine. Power Users Run It Like a Workforce.

This is the cheatsheet to the Patreon video. The 5 prompts inside are the exact moves I used to spin up a virtual data science team for one of my federal customers. Real project. Real output. About an hour.

What you get: The 5 prompts in order, with placeholders you can swap for your business, your project, and your role. Each prompt also includes the why, when to use it, and a power tip you can apply this week.

Who this is for: Operators, founders, consultants, and leaders running real businesses who want AI to act like a team, not a chatbot. You do not need to be a developer. You do need to be willing to think in workflows instead of one-off questions.

How to use it: Pick a real project. Open Claude Desktop. Walk through the 5 prompts in order. Do not skip step 1. The order is the leverage.

Important
These prompts work best in Claude Desktop with a paid plan, pointed at a folder on your computer. That keeps your data local. Keep "share with Anthropic" turned off in settings if you handle customer data.
01
Move One

Onboard Before You Ask

Most people open Claude and dump a question on it. That is the equivalent of hiring a consultant and asking for a report on day one. They have no context. Move 1 is to onboard the AI before you give it work. Two minutes of reading on the front end buys you hours of accuracy on the back end.
The Prompt
Read every file in this folder, including [any past performance, contract, or reference docs], so you have full context on what we are doing for [customer or project name]. Do not start any work yet. When you are done reading, tell me you are ready and give me a brief summary of what you understood. I have a real task for you after you confirm you are oriented.
Power Tip
Put everything related to the project in one folder: contract, past performance, raw data, key emails, prior reports. The folder is the AI's context. The richer the folder, the smarter the answers.
02
Move Two

Promote It from Generalist to Specialist

Generic prompts give you generic answers. Once Claude is up to speed, give it a role before you give it a task. Specialists outperform generalists every time. The same prompt, run by a "data scientist" instead of "Claude," produces a noticeably sharper output.
The Prompt
You are now a high-end, full-blown [specialist role: e.g., data scientist, financial analyst, capture manager] with the ability to [core capability: e.g., read raw data and build custom reports]. Focus on the data and docs in [folder or file name]. Here is what I need: 1. [Specific question one] 2. [Specific question two] 3. [Specific question three] Give me the answer in a way I can [present to my customer / share with my team / file with my CFO]. Also tell me anything I am not asking for but that you think I should know. Take your time and walk through it.
Power Tip
Use voice-to-text to dictate the prompt. When you talk it out, you give way more context than when you type. The richer the prompt, the better the work.
03
Move Three

Build a TEAM That Talks to Each Other

This is the leap from chatbot to workforce. Instead of one AI working alone, spin up a team of specialist agents and make them collaborate. Round 1: each agent does its own analysis. Round 2: they review each other's work. The lead synthesizes. The QA validates. Then you see results.
The Prompt
Before you produce any deliverables, I want you to build a [domain: e.g., data science, capture, finance] team. Now that you know what I am asking for, what would a team of specialists look like for this project? Build out the team. Each member is its own agent. Suggest the right roles based on the work (for example: [data engineer, analyst, QA validator, team lead]). Then make sure they talk to each other. Round 1: each agent does their own analysis. Round 2: they review each other's findings and cross-pollinate. There needs to be a team lead who runs the stand-up, and a QA validator who double-checks the numbers before you show me a single result. Bring me fresh findings as a team, not as individuals.
Power Tip
If your project is important, build a QA validator into the team. The QA layer catches the small errors that one-pass AI work tends to miss. It is the difference between AI that demos and AI that ships.
04
Move Four

Save the Institutional Knowledge

Most people forget this and pay for it later. After every important session, save the role, the team, the methodology, and the findings to a file inside the project folder. Next week you do not start from scratch. You pick up where you left off.
The Prompt
Do not let me lose this approach. Create or update a CLAUDE.md file inside this project folder that captures: 1. The customer or project context 2. The role you played (for example: [data scientist team lead]) 3. The agent team structure and what each member does 4. The methodology you used to produce these findings 5. The key findings themselves 6. Anything I should remember next time I open this folder Next time I open this folder, you should read this file first and pick up exactly where we left off. Do not skip anything important. Make it complete.
Power Tip
A well-written CLAUDE.md file is like an employee with perfect memory who never quits, never takes vacation, and never bills you for ramp-up. That one file compounds your leverage every single week.
05
Move Five

Branded, Ready-to-Ship Deliverables

The output is what your customer sees. Tell Claude to build the deliverables in parallel and brand them with your company colors and fonts. This is the difference between AI that demos well and AI that ships.
The Prompt
Now build me three deliverables in parallel: 1. A PowerPoint deck with the executive findings 2. A Word document with the full report 3. An Excel spreadsheet with the underlying data Use my brand standards ([colors, fonts, logo placement]) so all three look like they came from [your company]. World-class graphics. Clean executive layout. Ready to send to [customer name or stakeholder]. When you are done, save everything we learned today to the CLAUDE.md so we do not start from scratch later. I want all three deliverables ready in parallel, not one at a time.
Power Tip
Build a brand skill in Claude that captures your colors, fonts, and logo standards once. Every future deliverable will use it automatically. Set it up one time, benefit forever.
The Full Sequence

The 5 Moves Together

Run them in order. The order is the leverage. Skipping moves 1 and 2 is why most people get mediocre results.

01
Onboard Point Claude at the project folder. Tell it to read everything. Wait for confirmation.
02
Specialize Promote it to a specialist role. Give it the task and ask for what you didn't think to ask.
03
Build the Team Spin up a multi-agent team. Make them talk to each other. Add a lead and a QA.
04
Save the Knowledge Lock the role, team, methodology, and findings into a CLAUDE.md inside the folder.
05
Ship the Deliverables Generate PowerPoint, Word, and Excel in parallel. Branded. Customer-ready.
Before You Run It

Quick Checklist

Run this once before you start your first 5-move session. Set up takes about 10 minutes. Pays back in hours.

Claude Desktop installed with a paid plan (Pro or Max).
One folder per project on your computer with all relevant files inside.
"Share with Anthropic" turned off in settings if you handle customer or sensitive data.
Voice-to-text ready (built into Mac and Windows). Use it to dictate prompts so you give richer context.
Brand assets in one place: your colors, fonts, and logo standards. Save once, use forever.
30 to 60 minutes blocked for your first run. After that, you will move much faster.
One real project picked. Do not practice on a fake task. Use a real one. The leverage is in the realness.
Want More?

This Is Just One Workflow

I run plays like this every week across my businesses. If you want the next one, the live walkthroughs, and the prompt library that grows every month, come join me on Patreon.

Join the Patreon  →
By Rick Pina
Patreon.com/RickPina