AI Tools

Sample Simple Prompts 
Knowing what to ask the AI is the key to unlocking its power. A good prompt is specific, provides context, and clearly defines the type of answer you want. 

To get you started, here are some real-world examples from different departments using Apple as the example company. Feel free to copy, paste, and modify them for your own needs.    

For Marketing  

• Social Media: ”Write a LinkedIn post announcing the new iPhone 15 Pro. Focus on its titanium design, A17 Pro chip performance, and advanced camera system for professional content creators.”  
• Content Ideas: ”Generate 5 blog post titles about how Apple ecosystem integration improves productivity for remote teams.” 
• Ad Copy: ”Create three short, catchy headlines for a Google Ad promoting the MacBook Air with M4 to college students.” 

For Sales 

• Email Drafting: ”Draft a follow-up email to a potential client I met at CES. They were interested in deploying iPads for their retail staff. Keep the tone professional and suggest a short demo call next week.” 
• Meeting Prep: ”I have a meeting with the IT manager of a large retailer. Summarize the top 3 advantages of deploying iPhones with Mobile Device Management (MDM) for secure enterprise use.” 
• Role-Playing: ”Act as an enterprise buyer unsure about switching from Windows laptops to Macs. I’ll try to address your concerns. Start by explaining your main hesitation.”   

For Technical Support & Engineering  

• Simplifying Concepts: ”Explain how Apple’s Neural Engine enhances on-device AI features in simple terms suitable for a non-technical audience.” 
• Troubleshooting: ”Create a step-by-step troubleshooting guide for an iPhone that won’t connect to Wi-Fi. Include both software and router-related causes.” 
• Code Generation: ”Write a Swift function using CoreBluetooth to scan for nearby devices and display their names in the console.”  

Your Turn: Pick one of these prompts, test it with your AI assistant, and then tailor it for a task you’re working on this week. 

AI Prompting Cheat Sheet 
Prompting is not about asking AI a question – it’s about communicating intent clearly.  

The success of a prompt relies on the user providing clarity:   

• Define the role (persona): Tell the AI who it is to narrow its focus. 
• Provide all the needed facts (context): Give it all the information it needs, assuming it knows nothing.  
• Specify structure and tone (format): Dictate exactly how you want the output to look and sound. 
• Demonstrate examples (few-shot): Show, don’t just tell, what a good result looks like. 
• Encourage reasoning (Chain of Thought [CoT] and Tree of Thought [ToT]): Ask it to think step-by-step or explore multiple solution paths. 
• Review the process critically: A bad AI response is a sign that instructions or context need more clarity.

Provide Context & Verify the Output
The single most important skill for getting truly exceptional results with LLMs is prompt engineering. It boils down to two golden rules: 

• Rule 1: Provide Context
• Rule 2: Verify the Output

Think of AI as a brilliant, incredibly fast intern. It has a vast range of knowledge, but it doesn’t know your business, your clients, or your specific goals unless you tell it.  

Rule 1: Provide Context (Garbage In, Garbage Out) 
A vague prompt will get you a generic, often useless, answer. A detailed prompt with good context will get you a tailored, high-quality result.   \

Example: A Vague Prompt vs. a Context-Rich Prompt   

• Vague Prompt: ”Write a product description for the iPhone.” 
• Result: You’ll get a generic description based on public web data, which might be outdated or miss key selling points. 
• Context-Rich Prompt: “Act as a senior marketing strategist for Apple. Write a product description for the iPhone 15 Pro Max. Target Audience: Tech-savvy professionals who upgrade every two years. Key Features to Highlight: A17 Pro chip performance for productivity and gaming, titanium frame for durability and lightweight design, improved battery life, and advanced camera capabilities for content creation. Goal: Position the phone as a premium productivity and creative tool for on-the-go professionals.” 
• Result: You’ll get a detailed, high-impact description that resonates with the target audience’s priorities — performance, design, and usability — while highlighting the most compelling differentiators.  

Rule 2: Verify the Output (Trust, but Verify)
LLMs are designed to generate plausible text, not to be a source of absolute truth. They can occasionally “hallucinate” or invent facts, statistics, or technical details. 

Your Action Plan:   

• For Facts & Figures: If the AI gives you a statistic, a market size, or a competitor’s price, always double-check it with a reliable source (e.g. company internal data, a trusted industry report). 
• For Technical Details: If you ask for a code snippet or a hardware specification, always test it or cross-reference it with official company documentation. 
• For Creative Work: For emails or marketing copy, always review and edit it to ensure it matches the company’s tone and brand voice.  

By giving LLMs clear context and always verifying its output, you transform it from a simple chatbot into a powerful tool for brainstorming, drafting, and accelerating your work. 

Few-Shot Prompting: Teaching AI with Examples 
Few-shot prompting is a method where you show an AI a few examples of what you want — both the input and the output — before asking it to create something new. This helps the model understand your format, tone, and intent, so it produces results that closely match your expectations.  

Writing Sales Follow-Up Emails
Instead of simply asking, “Write a follow-up email to a potential customer,” you can show the AI a few samples first: 

Example 1
Input: Follow up after an initial demo with a positive lead. 

Output:
Hi [Name],  

I hope you enjoyed yesterday’s demo! As mentioned, I’m sending over more details about the [Product Name] features your team liked most. Let me know if you’d like to schedule a deeper dive next week.  

Best, 
Jake

Example 2
Input: Follow up after sending a proposal but receiving no reply.

Output: 
Hi [Name],  

Just wanted to check in about the proposal we shared last week. If you have any questions or feedback, I’d be happy to go over the details. When would be a good time for a quick catch-up?  

Thanks,
Jake 

Final Prompt Input: Follow up after a trade show conversation where the lead showed interest but hasn’t replied yet. 

By seeing these examples, the AI understands:  

• The tone should be professional yet friendly.
• The structure includes greeting, context, and a call to action. 
• The message is short and conversational.  

When you supply the new situation (trade show lead), the AI continues in the same format, producing a consistent, high-quality email.  

Why This Works
Few-shot prompting shows, rather than tells, what you want. It helps the AI learn:  

• Style and tone: friendly, concise, professional
• Structure: intro > context > next step
• Consistency: same formatting across all outputs

Few-shot prompting is like giving your assistant a few good examples before asking them to write something new. It’s especially useful for structured or stylistic tasks such as sales emails, marketing copy, or customer service replies. By setting the pattern upfront, you shape the results to fit your needs with minimal editing later. 

Using Personas
Assigning AI a specific expert role for a single task is a technique is called “Persona Prompting,” and it’s what turns a generic response into advice from a virtual specialist. By telling the AI who to be, you narrow its focus, improve its reasoning, and get more relevant, high-quality results.  

The “Act As…” Framework
The easiest way to use this technique is to start your prompt with: “Act as a [Role]…” This immediately signals the AI to adopt the mindset, tone, and skillset of that role for the task at hand. 

Examples Across Departments
Marketing – Competitive Analysis  

• Vague Prompt: “What are some weaknesses of Samsung phones?” 
• Persona Prompt: “Act as a Market Intelligence Analyst specializing in the smartphone industry. I need a competitive analysis of Samsung’s Galaxy S24 lineup. Focus on identifying potential weaknesses in performance, pricing, ecosystem integration, or user experience that Apple could emphasize in iPhone marketing. Present the insights as a bulleted list of ‘Competitive Gaps.’”  

Sales – Quarterly Business Review Preparation  

• Vague Prompt: “Help me get ready for my QBR with a reseller.” 
• Persona Prompt: “Act as a Senior Partner Account Manager. I’m preparing for a quarterly review with a major reseller of Apple devices. Create a presentation outline that summarizes joint wins from the last quarter, identifies growth opportunities in education and enterprise, and proposes three new co-marketing initiatives for next quarter.”   

Technical Support – Writing Documentation 

• Vague Prompt: “Explain how to connect an iPhone to Wi-Fi.” 
• Persona Prompt: “Act as a Technical Writer creating a support article for Apple’s Help Center. Write a clear, step-by-step guide on how to connect an iPhone to a Wi-Fi network. The target audience is non-technical users. Include numbered steps, troubleshooting tips, and a short ‘Common Problems’ section.” 

Operations – Process Improvement  

• Vague Prompt: “How can we speed up our device repair process?”
• Persona Prompt: “Act as a Lean Six Sigma Consultant. Our current iPhone repair process averages 12 business days from intake to customer delivery. Review the following process steps [paste steps here]. Identify the top three bottlenecks and recommend concrete improvements to cut turnaround time by at least 30%. Focus on logistics, queue management, and communication flow.”  

Try It Yourself: Next time you use any AI assistant, don’t just ask a question — assign it a role first. Start with “Act as a…” and watch how the depth, precision, and usefulness of its responses dramatically improve. 

For additional details, see The Ultimate Ideal Persona Generator

C.R.A.F.T. Framework
A prompt isn’t just a question — it’s a set of instructions. The clearer and more complete your instructions are, the better the AI’s output will be. To make this simple to remember, use the C.R.A.F.T. framework. 

C – Context: “What you need to know.”  

• Provide background, data, and any relevant details so the AI fully understands the situation. Don’t make it guess. 
• Example: “We’re preparing a launch campaign for the iPhone 15 Pro. Our main competitors are Samsung and Google. Here’s our internal spec sheet: [paste data].” 

R – Role: “Who you need to be.”  

• Assign the AI an expert persona. This is the “Act as…” technique focuses the AI’s knowledge, tone, and reasoning. 
• Example: “Act as a senior technology marketing strategist specializing in smartphone launches.” 

A – Action: “What I need you to do.”  

• Be clear about the task. Use direct verbs such as analyze, create, summarize, or write.
• Example: “Create three distinct positioning angles for the iPhone 15 Pro launch campaign.” 

F – Format: “How I want it to look.”  

• Specify the desired output structure — bullets, table, JSON (JavaScript Object Notation), email draft, report, etc. 
• Example: “Present the output as a three-column markdown table labeled ‘Angle,’ ‘Key Message,’ and ‘Target Audience.’” 

T – Tone: “How it should sound.”  

• Define the voice and style — professional, conversational, persuasive, technical, or friendly. 
• Example: “The tone should be confident, premium, and aspirational, reinforcing Apple’s brand identity.” 

Putting It All Together  
Before (Simple Prompt): “Write some marketing slogans for the iPhone 15 Pro.”  

After (C.R.A.F.T. Prompt):  

• [R] Act as a senior technology marketing strategist specializing in smartphone launches.  
• [C] We’re preparing a campaign for the iPhone 15 Pro, which introduces a titanium frame, A17 Pro chip, and professional-grade camera system. Competitors include Samsung Galaxy S24 and Google Pixel 8 Pro. 
• [A] Create three unique positioning angles for the launch marketing campaign. 
• [F] Present the output as a markdown table with columns: “Angle,” “Key Message,” and “Target Audience.” 
• [T] The tone should be confident, premium, and aspirational.  

Before sending a prompt, run a quick mental check against the C.R.A.F.T. framework. You won’t always need every element, but the more you include, the more accurate, creative, and ready-to-use your first result will be. 

Chain-of-Thought (CoT) vs. Tree-of-Thought (ToT) Prompting 
When you need higher-quality reasoning from an AI — especially for analysis, troubleshooting, planning, or creative problem-solving — explicitly guiding how it thinks can dramatically improve accuracy. 

Two widely used methods are Chain-of-Thought (CoT) and Tree-of-Thought (ToT) prompting. Modern AI systems already use similar internal reasoning patterns, but explicitly prompting the model to use them helps align its reasoning process with your expectations and objectives.  

Chain-of-Thought (CoT) Prompting 
Definition: Chain-of-Thought prompting is the practice of instructing the AI to “think step-by-step.” Instead of providing a final answer immediately, the AI lays out its reasoning process, explaining each stage of how it gets there. 

Why It Works: LLMs are trained to predict text sequences. When you ask them to reason step-by-step, you guide that process toward slower, more deliberate reasoning paths. This significantly improves accuracy for logical, mathematical, diagnostic, and multi-step procedural tasks. 

Best For:  

• Debugging technical issues  
• Multi-step data or financial analysis  
• Planning workflows or processes 
• Justifying decisions or recommendations  

How to Use It: Include directives like:  

• “Let’s reason this through step-by-step.” 
• “Show your reasoning for each stage before concluding.” 
• “Explain your thought process before giving the final answer.” 

CoT Example 
Prompt: “Act as a systems engineer. Diagnose why this network share intermittently disconnects. Think step-by-step: list potential causes, test each hypothesis logically, and then recommend the most likely root cause and fix.”  

Tree-of-Thought (ToT) Prompting
Definition: Tree-of-Thought prompting expands on CoT by teaching the AI to reason through multiple possible paths before deciding on the best one. Instead of a single chain of reasoning, the AI explores alternatives, like a decision tree, evaluates trade-offs, and converges on the most effective solution. 

Why It Works: ToT takes advantage of the model’s ability to simulate branching reasoning, like how humans explore options before committing. It increases creativity and decision quality, especially for open-ended or strategic problems.  

How to Use It: Guide the AI to branch and evaluate:  

• “List several possible solutions and evaluate each.”  
• “Identify two or three potential approaches, weigh pros and cons, then recommend the best one.” 
• “Explore multiple paths before finalizing your answer.” 

Best For:  

• Product or business strategy 
• Brainstorming and innovation 
• Architecture and design trade-offs
• Evaluating competing ideas or plans 

ToT Example 
Prompt: “Act as a product manager. We need to boost user adoption of our new iCloud feature. First, brainstorm three strategic directions (Tree-of-Thought), then evaluate each for effort, impact, and risk. Finally, recommend the optimal approach and justify your decision.” 

How Modern “Thinking” Models Incorporate This
Recent generations of LLMs (ChatGPT-4, Claude 3, Gemini, and others) already integrate structured reasoning internally, essentially performing hidden Chain-of-Thought-style processing before generating an answer. 

Research models and scaffolding systems (e.g. Reason+Act, Reflexion, and Tree-of-Thought agents) simulate branching logic and self-evaluation internally.  

That means you don’t always need to over-specify each step but prompting explicitly still makes a measurable difference. When you say “think step-by-step” or “explore multiple possible paths,” you help the model externalize this reasoning in a transparent, controllable form.