Working Smarter with Claude · Purium Team Training

01

The Five Prompting Philosophies

Most people use Claude the way they used Google: type a quick question and hope for the best. That's one way to use it. It's also the least powerful way. There's a full spectrum of approaches, each unlocking a deeper level of collaboration. Understanding where you are on this spectrum is the first step to leveling up.

Basic

Mastery

🔍 Search Engine Mode

How most people start

Quick one-liner questions. Good for simple lookups. Output is generic because input is generic.

"What is chlorophyll good for?"

⚙️ Prompt Engineering

Structure and specificity

Deliberate, formatted prompts with role, task, and constraints. Output quality rises significantly.

"You are a wellness copywriter. Write 3 Instagram captions about the Power Shake. Tone: warm, energizing. Under 120 chars each. No hashtags."

🧠 Context Engineering

Richness before the task

Load Claude with everything it needs to know before making a request. Brand voice, audience, history, product details. The heavier the context, the more tailored the output.

"[paste Brand Voice doc] [paste Product Sheet] Now write a launch sequence for the Ultimate Lifestyle Transformation 12-Week program."

🎯 Outcome Engineering

Start with the destination

Define what success looks like, then work backwards. Tell Claude what a great output would accomplish, not just what it should contain.

"Write a Brand Partner welcome email. The reader should feel proud, excited, and clear about their first 3 actions. If they finish reading and feel confused, it failed."

💡 Intent Engineering

The most advanced approach

Communicate the WHY behind the request so Claude can make better judgment calls throughout. Especially powerful for complex, multi-part work.

"I am the Director of Purium. I need to brief our dev team on a shopping site update. My goal is clarity, not technicality. They should know exactly what to build and why it matters to the customer experience. Use that as your lens."

Key Insight

You don't have to choose one. The best prompts stack these philosophies: rich context (Context Engineering) + a defined role (Prompt Engineering) + a clear success measure (Outcome Engineering) + a stated purpose (Intent Engineering). That's the masterclass in one prompt.

The R-C-T-F-C Formula: Prompt Engineering Made Simple

Your Everyday Prompting Backbone

Role + Context + Task + Format + Constraints = Consistent, High-Quality Output

R
Role: "You are a Purium Brand Partner coach specializing in onboarding new members..." — Sets vocabulary, judgment, and tone in one line.
C
Context: Who is the audience? What's the situation? What has already been decided? Don't make Claude guess what you know about Purium.
T
Task: "Write three email subject lines for our Ultimate Lifestyle Transformation program" beats "help with marketing."
F
Format: "Give me three numbered options, under 60 characters each." Prevents reformatting work later.
C
Constraints: "Do not make any health claims. Do not use the word 'detox' without qualification. Always include our FDA disclaimer language."

Before vs. After: Purium Style

Before: Search Engine Mode

Write a post about the Power Shake.

After: Prompt + Context + Outcome Engineering

You are a wellness copywriter for Purium, "The Transformation Company." Our Power Shake (Apple Berry) is our most popular product — known as "Liquid Sunshine." Our audience: health-conscious adults seeking clean, organic nutrition without processed food. Write 3 Instagram captions for the Power Shake. Each should make the reader feel: energized and hopeful. Tone: warm, knowledgeable, non-pushy. Under 130 characters. No health claims. No hashtags in body.

Three Habits to Build Starting Today

Give Claude a role first. Start with "You are a [role] for Purium..." and every output immediately gets better.
Iterate, don't restart. If the output is close: "This is good. Now make it warmer, shorten by 20%, and move the CTA to the second paragraph." Claude holds the full conversation context.
Paste examples. "Write in the style of this email I love: [paste]." Claude will match cadence, structure, and feel.

Power Move: Turn It Around and Ask Claude What It Needs

One of the most effective prompting techniques is letting Claude tell you what it's missing before it starts. After giving your initial brief, ask something like:

"What else would you like to ask me to make this the most optimal and effective?"

or, when launching a complex task:

"What questions do you have about this request to execute it at a high level?"

This is the exact prompt used to turn a Zoom transcript into a full Purium branded webpage (see the Content Creation use case below). Claude asked clarifying questions, the brief got sharper, and the output needed far less editing. Make this a habit for any high-stakes or complex task.

Purium Compliance Note

Purium sells supplements and superfoods. All copy must include FTC/FDA-compliant language. Always instruct Claude to follow these rules in your prompts. Make this part of your Project instructions so it fires automatically.

Include in every Project instruction: "All copy must comply with FTC and FDA guidelines. Do not make disease treatment or cure claims. Always append: 'These statements have not been evaluated by the Food and Drug Administration. This product is not intended to diagnose, treat, cure, or prevent any disease.' when applicable."

Try It Now

Open Claude. Type the weak prompt above and copy the output. Then type the strong prompt and compare. Notice how specificity drives quality.

02

Building Your Brand Documents

The single biggest unlock for getting consistent, on-brand Claude output is not better prompting: it's better context files. When you upload a Brand Voice document or a Brand Guidelines reference into a Claude Project, Claude reads it before every conversation inside that project. You stop re-explaining. Claude already knows.

These are the documents worth creating now. Each one takes 30-60 minutes to build and saves hours every week.

🎙️

Brand Voice Document

How Purium sounds in writing (3-5 adjectives + explanation)
What the brand never says or sounds like
Tone shifts by platform (Instagram vs. email vs. formal letter)
Power words Purium uses (transformation, pure, premium, nourish, thrive)
Words to avoid ("detox" without context, disease claims, hype language)
2-3 examples of ideal Purium copy with annotations

Give this to Claude and say: "Always write in this voice."

📐

Brand Guidelines Document

Mission: "Pure + Premium" nutrition for transformation
Core values (6 Purium core values)
Visual direction notes (color palette, logo usage rules)
Product naming conventions (exact spelling, capitalization)
Platform-specific formatting rules
Legal/compliance requirements per content type

Especially useful for anyone creating on-brand assets or presentations.

🌿

Product Reference Sheet

Full product list with correct names and capitalization
Key benefits per product (Power Shake, Biome Medic, Super Amino 23, Sleep Nightly, Alpha Omega, etc.)
Program structures (Ultimate Lifestyle Transformation 12-Week, LTD Lifestyle Transformation Daily)
Approved benefit language per product
Ingredients to highlight per product
Claims to never make per product

Upload this to every marketing and Brand Partner project.

👥

Audience Persona Document

Primary customer persona (health-conscious adult, 35-55)
Brand Partner persona (values-aligned, entrepreneurial)
Pain points for each audience
Language each audience uses about their goals
Objections each audience has before buying
What makes each persona say yes

Paste this into any prompt where audience matters.

⚖️

Compliance + Claims Guide

FDA disclaimer (full, approved text)
List of approved structure/function claims
Prohibited disease claims by product category
Testimonial usage rules
Income disclosure requirements for Brand Partner content
Platform-specific restrictions (Facebook, Instagram, email)

This is non-negotiable. Every AI-generated copy must pass this filter.

🤝

Brand Partner Program Guide

Program tiers and titles (Diamond, Black Diamond, Red Diamond, Crown)
Compensation structure overview
Approved language for income opportunity discussions
Onboarding steps and resources
Support resources Brand Partners have access to
Key messaging for recruiting vs. product sharing

Essential for any Brand Partner communications or training materials.

Team Activity

In pairs, pick one document (2 min): Which of these six does your role need most urgently? What information do you already have that would go into it? What's missing? You'll start building this in the practice round.

How to Build These Documents Using Claude

You don't have to write these from scratch. Claude can help you create them. Use this to get started:

You are a brand strategist helping Purium Health Products, "The Transformation Company," create its internal Brand Voice document. Purium = Pure + Premium. We sell organic superfoods and supplements. Our founders are Dave Sandoval and Amy Venner. Our mission is transformation: clean nutrition that improves health, energy, and wellbeing. Interview me with 8 focused questions to gather what you need to build a complete Brand Voice document. Then produce a final document I can save, formatted for upload into Claude Projects.

03

Projects: Claude's Long-Term Memory

By default, Claude remembers nothing between conversations. Every new chat is a blank slate. If you've been pasting your brand context or product details every single time, that's the cost of not using Projects. A Project is a persistent workspace where Claude always knows your context, your files, and your preferences.

What Lives Inside a Project

1
Instructions (System Prompt): What Claude should always know before any conversation inside this project. Your brand voice, audience, compliance rules, and formatting preferences. Set it once; it fires every time.
2
Knowledge Files: Upload your Brand Voice doc, Product Reference Sheet, Compliance Guide, Brand Partner Program overview. Claude reads all of them and references them throughout every conversation.
3
Persistent Context: Chats inside the project share history. Refer back to previous decisions without re-explaining them.

Sample Project Instructions: Purium Marketing

You are the AI marketing collaborator for Purium Health Products, "The Transformation Company." Brand: Pure + Premium organic superfoods. Founded by Dave Sandoval and Amy Venner. 25+ years in clean nutrition. Voice: Warm, knowledgeable, empowering. We inspire action without pressure. We celebrate transformation, not perfection. Audience: Health-conscious adults 35-55 seeking clean nutrition, energy, and whole-body wellness. Also: values-aligned entrepreneurs exploring our Brand Partner opportunity. Tone never to use: Hype language, aggressive sales pressure, MLM stereotypes, disease cure claims. Compliance: ALL copy must comply with FDA/FTC guidelines. No disease treatment claims. Always flag when FDA disclaimer is needed. Do not use "cure," "treat," or "prevent" in reference to products. Products to know: Power Shake (Apple Berry = "Liquid Sunshine"), Biome Medic, Super Amino 23, Super CleansR, Sleep Nightly, Alpha Omega. Programs: Ultimate Lifestyle Transformation (ULT, 12-Week), LTD (Lifestyle Transformation Daily). Format: Always provide copy in numbered variations unless asked otherwise. Ask clarifying questions before long tasks if the brief is vague.

Purium Project Ideas for Your Team

📧

Email Campaigns

Transformation launches, Brand Partner updates, nurture sequences

📱

Social Content

Instagram, Facebook, captions for Power Shake, ULT, and lifestyle content

🤝

Brand Partner Training

Onboarding curriculum, scripts, talking points, FAQ documents

🖥️

Site Dev Briefs

Developer instructions for ishoppurium.com updates and features

📝

Executive Comms

Letters, memos, curriculum documents, partner announcements

🎥

Zoom → Content

Paste a meeting transcript and turn it into webpages, emails, or social content

📊

Strategy + Research

Market research, competitive intel, audience analysis, trend reports

Try It Now

In Claude, click "Projects" in the left sidebar. Create a project called "Purium Marketing." Paste the system prompt above into the Instructions field. Start a conversation inside the project asking for a Power Shake Instagram caption. Then open a fresh chat outside the project and ask the same thing. The difference will be immediate.

Other Corporate Use Cases

These prompt structures support team members working on curriculum, technical communications, and formal correspondence.

📚 Training Curriculum

You are an instructional designer for Purium. Create a [X]-module training curriculum for [audience: new Brand Partners / customer service staff]. Each module should include: learning objectives, key content, an activity, and a short quiz. Tone: motivating, clear, and grounded in Purium's values. Format as a structured outline I can expand.

🖥️ Developer Briefs

You are a product manager writing a developer brief for the ishoppurium.com team. The update needed is: [describe change]. The goal from a customer experience perspective is: [goal]. Write a clear, non-technical summary first, then a technical specification with: what to build, acceptance criteria, and edge cases to handle.

✉️ Letters + Comms

You are drafting a letter on behalf of [Name] at Purium. Recipient: [Brand Partners / corporate partner / vendor]. Purpose: [announce / acknowledge / clarify / thank]. Tone: professional, warm, and aligned with Purium's values. Length: [1 page / under 300 words]. Include a clear next step or call to action.

Corporate Project Tip

Create a separate "Executive Comms" or "Corporate" project with the sender's name, title, signature preferences, Purium's formal voice, and any letter or document templates already uploaded. Claude will then draft everything in that voice automatically, significantly reducing review and edit time.

Marketing Use Cases: Ready to Copy

Email Campaign

Transformation Program Launch

You are a conversion copywriter for Purium. Write a 3-email launch sequence for the Ultimate Lifestyle Transformation 12-Week program. Email 1: curiosity + the problem. Email 2: the solution (product + program). Email 3: urgency + testimonial hook. 150-180 words each. Warm, motivating tone. Comply with FDA guidelines. No disease claims.

Social Media

Repurpose Blog to Social

Here is a Purium blog post: [paste]. Repurpose it into: 1 Facebook post (200 words), 3 Instagram captions (under 130 chars each), 5 one-liner tweets. Keep Purium's warm, knowledgeable tone. No disease claims. Include FDA disclaimer language where relevant.

Brand Partner

New Partner Welcome

You are a Purium Brand Partner success coach. Write a welcome email for a newly enrolled Brand Partner. They should feel: proud, clear on their first 3 steps, and part of a mission larger than selling. Tone: warm, community-oriented, not salesy. 200 words. Include income disclosure reminder.

Content Creation

Zoom Transcript → Webpage

Please create a Purium branded webpage from this meeting and ensure it is compelling, engaging, and actionable, as well as FTC compliant. It should be exciting, encouraging, and informative. This is for the marketing of Purium products to Customers and positioning for Brand Partners. The main goal is a high level of conversion. What questions do you have about this request to execute it at a high level? [Paste your Zoom transcript here]

04

Claude Models: Choosing the Right Tool

Not all versions of Claude are the same. Anthropic offers a family of models, each built for different levels of complexity and speed. Think of it like Purium's programs: LTD handles your everyday nutritional foundation; the 12-Week ULT is for a deeper transformation. Each serves a different need.

Fastest

Claude Haiku

Quick, lightweight, low-cost. Best for simple, repetitive tasks where speed matters more than depth.

Quick subject line suggestions
Simple formatting tasks
Lookup-style questions
High-volume, templated outputs
Chatbot or customer FAQ responses

LTD of AI models

⭐ Best for Most Work

Claude Sonnet

The everyday workhorse. Strong reasoning, excellent writing, fast enough for real-time use. This is the model you'll use most.

Marketing copy and campaigns
Brand Partner communications
Email sequences
Social content creation
Training curriculum drafts
Developer briefs and letters

Your everyday ULT of AI

Most Capable

Claude Opus

Deep reasoning, complex analysis, nuanced judgment. Use when the task is genuinely complex or high-stakes.

Strategic planning documents
Complex multi-part analysis
Legal or compliance review support
Long-form curriculum design
Sensitive executive communications
Research synthesis across many sources

Your 12-Week ULT of AI models

Where to Switch Models

Inside Claude.ai, click the model name in the top-left corner of any chat to switch. For most of what we do at Purium, Claude Sonnet is the right default. Move to Opus when a task requires deep reasoning or complex structure. Use Haiku only when speed and volume are the priority over nuance.

Important Note on Model Names

Anthropic updates its model family regularly. The model you see in the interface may have a version number (e.g., "Claude Sonnet 4"). When using the API for any automated workflows, always use the exact model string provided in Anthropic's documentation to avoid unexpected behavior.

05

Skills: Extending Claude's Capabilities

Skills (also called integrations or tools) are add-ons that let Claude take actions beyond conversation: searching the web, pulling live data, connecting to apps, running code. They are powerful. They also carry real risk when they come from unknown sources.

Non-Negotiable: Skills Safety at Purium

Third-party Skills can request access to your Claude conversations, your files, and your connected accounts. An unknown Skill could be designed to exfiltrate business-sensitive data or inject hidden instructions into Claude that you never see.

Third-party Skills may have access to everything you type in Claude
Malicious Skills can execute "prompt injection" — hidden instructions that override yours
Skills with access to our systems could expose customer data or Brand Partner information
A compromised Skill could post, send, or modify things on your behalf

The rule is simple: only install Skills verified by Anthropic, or explicitly approved by leadership. When someone sends you a link to install a Skill — ask first. No exceptions.

Safe, Built-In Capabilities (No Third-Party Risk)

These come from Anthropic directly and are safe to use in Claude.ai:

🔍

Web Search

Current market trends, competitor research, ingredient science, health and wellness news

📄

File Analysis

Upload PDFs, product sheets, spreadsheets, images; Claude reads and works with them

💻

Code Execution

Build data summaries, charts, process reports, automate repetitive text work

🗂️

Memory

Opt-in feature: Claude remembers your preferences across conversations

Before Installing Any Third-Party Skill, Ask Three Questions

(1) Is this made by Anthropic or a verified partner?
(2) What data is it requesting access to?
(3) Has leadership approved this for Purium team use?

If you cannot confidently answer all three: do not install it. Come ask first.

06

Evaluations: How Do You Know It's Actually Working?

Getting Claude to produce output is easy. Knowing whether that output is actually good, consistently, over time, across different team members, is a different skill entirely. That's what evaluations are for.

An evaluation (or "eval") is simply a structured way of asking: did this output meet the standard? It can be as simple as a checklist you run before publishing, or as sophisticated as a scoring rubric that multiple team members use to compare Claude's output across different prompt versions.

The Core Evaluation Question

Would a real Purium customer or Brand Partner read this and take the next step — or would they scroll past?

The Three Levels of Evaluation

1

Gut Check (30 seconds): Read it out loud. Does it sound like Purium? Does it feel warm, credible, and clear? Would you personally share this? If anything makes you hesitate, flag it before it ships.
2

Criteria Check (2 minutes): Run the output against a fixed set of standards: on-brand voice, correct product names, no discontinued items, compliance language present, CTA points to the right next step, no disease claims. This is where a checklist saves you.
3

Comparative Eval (ongoing): When you want to improve a prompt or template, run two versions against the same brief and score them side by side. Which one converts better? Which one sounds more like Purium? This is how you compound quality over time.

Ask Claude to Evaluate Its Own Output

One of the most powerful and underused techniques: after Claude produces a draft, ask it to evaluate the draft before you do. Claude will often catch its own weaknesses faster than you will.

Review the asset you just wrote against these criteria and score each from 1 to 5: 1. Brand voice — does it sound like Purium (warm, knowledgeable, empowering)? 2. Audience clarity — is it written for the right person at the right stage? 3. Conversion — does the CTA point to a clear, logical next step? 4. Compliance — are there any claims that could trigger FDA or FTC concern? 5. Product accuracy — are all product names current and correct? For any score below 4, explain what's weak and rewrite that section.

Evaluate a Prompt, Not Just an Output

If Claude consistently gives you mediocre results on a particular type of asset, the problem is usually the prompt, not the model. Use this technique to diagnose and improve:

Prompt Evaluation Exercise

Take a prompt that gave you output you weren't happy with. Paste it into Claude and say: "What is weak or missing in this prompt that might be causing mediocre output? Suggest a rewrite." Then run the improved version and compare. This is prompt engineering with Claude as your co-engineer.

The Purium Output Scorecard

A simple rubric your team can use consistently. Print it, paste it in your Project, or check it mentally before anything goes out.

✅

Green Light: Ship It

Sounds unmistakably like Purium
Written for the right audience at the right stage
CTA is clear and points to the logical next step
No disease claims, compliance language present
All product names current and correctly spelled
You'd personally share or send this

🔁

Red Flag: Revise Before Sending

Sounds generic, corporate, or over-hyped
Audience is unclear or mixed
CTA is vague ("learn more," "click here" with no context)
Contains a claim that could be read as a disease cure
References a discontinued product
You'd hesitate to put Purium's name on it

The Compound Effect

Every time you run an eval and improve a prompt, you raise the floor for every future output from that Project. This is exactly how Purium thinks about nutrition: consistent daily inputs that compound into transformation. Same principle applies to your AI workflow.

07

Hands-On Practice

Open Claude now. Choose the exercise that fits your role and work through it at your own pace.

Exercise A: Marketing Team — Prompting Upgrade

Take a prompt you've used before that gave mediocre results. Rebuild it using Context Engineering + the R-C-T-F-C formula. Add Purium brand context, product name, compliance instruction. Run both and compare.

Exercise B: Any Role — Build Your First Brand Document

Paste this into Claude: "Interview me to build a Purium Brand Voice document. Ask 6 questions, one at a time. Then write the final document formatted for upload into Claude Projects." Answer the questions. Save the output.

Exercise C: Corporate — Project Setup

Create a project called "Purium Corporate Comms." In the instructions, describe your role, your communication style, who you typically write to (Brand Partners, vendors, staff), and your preferred letter format. Then ask Claude to draft a one-page letter announcing a new company initiative of your choosing.

Exercise D: Zoom Transcript → Content

Find any Purium Zoom recording transcript (or paste a few paragraphs of meeting notes). Use this prompt: "Please create a Purium branded webpage from this meeting. Ensure it is compelling, engaging, and actionable, as well as FTC compliant. It should be exciting, encouraging, and informative. This is for the marketing of Purium products to Customers and positioning for Brand Partners. The main goal is a high level of conversion. What questions do you have about this request to execute it at a high level?" Answer Claude's questions, then review the output.

Exercise E: Any Role — Run an Eval

Take any output Claude has produced today. Paste the self-evaluation scorecard prompt above and ask Claude to score it. Note what it flags. Decide: ship it, or revise first?

08

Wrap-Up

The Four Things That Will Actually Change How You Work

1
Move up the prompting spectrum. The difference between Search Engine Mode and Intent Engineering is the difference between generic output and output that sounds like Purium. Invest 60 seconds more per prompt; save 20 minutes of editing.
2
Build your brand documents and upload them to Projects. Once you have a Brand Voice doc, a Product Reference Sheet, and a Compliance Guide inside your Claude Project, every output gets better automatically. This is the highest-leverage hour you can invest.
3
Ask Claude what it needs before it starts. On any complex task, end your brief with "What questions do you have to execute this at a high level?" You'll get sharper output with less back-and-forth.
4
Evaluate before you publish. Run the scorecard. Ask Claude to self-evaluate. Every output you improve raises the floor for the next one. This is where the compound effect kicks in.

Your First Action

Before your next piece of Purium content: open your Project (or create one), paste your brand instructions, use the full prompting formula, and run a quick eval before it ships. That's the whole system in one workflow.