Sponsored by

Read time: 6 minutes / Midjourney AI Prompt Library GOT A NEW UPDATE! If you bought The Complete AI Bundle* you can find the updates inside already (it will keep being updated for a few more weeks).

Greetings from above,

It's an exciting time in the world of AI, and I've got some juicy updates to share with you.

Claude, the AI assistant known for its natural, human-like tone, just dropped a new version that's causing quite a stir.

According to benchmarks, Claude 3 is even outperforming GPT-4, a feat no other model has achieved so far.

But as we all know, the real test is in the hands of the users. So, I've been putting Claude 3 through its paces, and I'm here to give you the scoop.

Today, we'll dive into:

• Why Claude wants you to use XML tags when prompting it

• 10 Prompt Engineering Tips for Claude 3

• Why the large context window is a game-changer

• A FREE Claude 3 Mega-Prompt Bundle (keep on reading)

Let's get into it!

Learn how to make AI work for you.

How do you stay up-to-date with the insane pace of AI? Join The Rundown – the world’s fastest-growing AI newsletter with over 500,000+ readers learning how to become more productive using AI every morning.

1. Our team spends all day researching and talking with industry experts.

2. We send you updates on the latest AI news and how to apply it in 5 minutes a day.

3. You learn how to become 2x more productive by leveraging AI.

Subscribe with one click.

⚙️ Why Claude Wants You to Use XML Tags

On Claude's official website, they recommend using XML tags when crafting your prompts.

The idea is that these tags help Claude better understand the different parts of your prompt, especially as your prompts grow in complexity.

I put this to the test with my one-shot prompt for validating my business ideas. By wrapping the instructions, context, and original asset sections in XML tags, I found that Claude was able to more accurately parse the information and deliver relevant results.

<role>You are a pragmatic business strategist with expertise in dissecting business ideas for real-world applicability.</role>

<task>Analyze the given business idea objectively, considering its genuine merits and potential pitfalls. Assume the roles of theoretical personas, offering realistic feedback on the idea's utility or lack thereof. Provide a blunt, balanced validation and recommendation.</task>

When responding, make sure sure to do so in the following <response_format>.

<response_format>
<business_idea_overview>$business_idea_overview</business_idea_overview>

<potential_markets>$potential_markets</potential_markets>

<persona1>
<age>$age</age>
<occupation>$occupation</occupation> 
<pain_points>$pain_points</pain_points>
<validation>"$problem. $solution_effectiveness $opinion"</validation>
</persona1>

<persona2>
<age>$age</age>
<occupation>$occupation</occupation>
<pain_points>$pain_points</pain_points>  
<validation>"$problem. $solution_effectiveness $opinion"</validation>
</persona2>

<persona3>
<age>$age</age>
<occupation>$occupation</occupation>
<pain_points>$pain_points</pain_points>
<validation>"$problem. $solution_effectiveness $opinion"</validation>  
</persona3>

<market_risks>$market_risks</market_risks>

<alternative_business_models>$alternative_business_models</alternative_business_models>

<final_validation_and_recommendation>$final_validation_and_recommendation</final_validation_and_recommendation>
</response_format>

<business_idea>
[PUT YOUR IDEA HERE]  
</business_idea>

GIVE OUTPUT WITHOUT XML TAGS!

Without the tags, large prompts can sometimes cause confusion, as I've experienced with ChatGPT.

But with Claude 3, the XML tags keep everything nicely organized, leading to more precise outputs.

⚙️ 10 Prompt Engineering Tips for Claude 3

1. Be Clear and Direct

Provide lots of context and clearly define the desired output. Use numbered lists, have a friend verify your prompts, and leave no room for ambiguity.

2. Show, Don't Tell with Examples

The magical "few-shot"PromptingTechnique: Provide 3-5 examples demonstrating the exact output you want, especially for structured tasks. Examples improve accuracy and consistency.

3. Give Claude a Role

Assigning a role like "academic" or "friendly tutor" helps Claude adapt its personality and vernacular for that situation.

4. Use XML Tags for Better Prompting

Claude's website recommends using XML tags when prompting to help it better understand the different parts, especially for larger prompts. For example:

<INSTRUCTIONS> Your instructions go here </INSTRUCTIONS> <CONTEXT> Provide context details here </CONTEXT> <ORIGINAL ASSET> Paste original asset like email, script etc. </ORIGINAL ASSET>

Using tags helps Claude separate instructions, context and source material, leading to more accurate results compared to unstructured prompts.

5. Chain Your Prompts

Break down complex tasks into sequential prompts, using the previous output as the next input. This improves focus and parallel processing.

Prompt 1: Extracting the quotes

Prompt 2: (using {{QUOTES}} output from Prompt 1): Answering the question:

6. Let Claude Think Step-by-Step

Have Claude reason through problems incrementally by thinking out loud before giving the final answer. Guides transparency too.

7. Prefill to Set the Tone

Prefilling provides starting text to steer Claude's language, formatting, persona, and more right from the get-go.

Here's an example of how to prefill Claude's response:

8. Control Output Format

Use examples, direct instructions, or prefilling to specify structured formats like JSON or code.

9. Ask for Rewrites

Have Claude refine its initial response when needed based on your feedback, instructions, and prompts.

10. Maximize the Long Context Window

Structure document prompts with XML tags, put documents upfront, have Claude reference relevant passages to reduce hallucinations.

These are just the basics - Anthropic's full prompt engineering cookbook is loaded with even more advanced strategies to truly master Claude.

Vision Capabilities: Extracting Text from Images

Claude's vision capabilities are pretty mind-blowing.

While there are likely many use cases yet to be discovered, I've been loving it for extracting text from images.

The Power of a Large Context Window

One of Claude 3's superpowers is its expansive context window. At launch, it can handle 200k tokens, with plans to extend to a whopping 1M tokens soon. For comparison, ChatGPT turbo sits at 128k tokens, while GPT-4's chat interface is limited to just 32k.

This large context window makes a world of difference. Not only can you upload and process huge amounts of data, but your chat conversations can go on much longer without losing any information.

🗂️ Free Resources For You 🗂️

Because I love you all I compiled a Mega-Prompt Bundle for Claude 3

that I am happy to share with you FOR FREE, no strings attached:

OFFICIAL Claude 3 Prompt Library:

28 ChatGPT Prompts to Help You Make Better Business Decisions [Cheatsheet]:

🧠 Did You Know? 🧠

By the way, my Midjourney AI Prompt Library GOT A MAJOR NEW UPDATE!

• Updated prompts, example outputs, how-to guides and tips & tricks

• More updates are going to be rolling out in the next few weeks

• If you bought The Complete AI Bundle, you can already see the updates inside.

• If you already bought the Midjourney AI Prompt Library before, you can already see the updates for free.

• If you buy it now, you will enjoy a lifetime access with updates FOREVER.

Wrap Up

Claude 3 is a seriously impressive update that's pushing the boundaries of what AI assistants can do. With its XML tag prompting, powerful vision capabilities, and expansive context window, it's a tool that every AI enthusiast should have in their toolkit.

I'm excited to keep exploring its potential and sharing my discoveries with you. I’m also working on making the biggest Claude 3 Mega-Prompt Bundle, so stay tuned for that.

Let me know if you have any other Claude 3 insights to share!

Keep learning,

🔑 God of Prompt