72 Hours with Claude Sonnet 3.5: What I Learned and Some Prompts for You

Welp, Anthropic dropped a bombshell this week…

So I scrapped the newsletter I was writing this week to instead share my explorations with their latest model in the past 72 hours.

But before we dive into my experiments, let's break down what's new with Claude 3.5 Sonnet:

The Big Updates

Beefed-Up Performance

Claude 3.5 Sonnet isn't just an incremental upgrade - it's flexing hard on the competition.

We're talking graduate-level reasoning, undergrad knowledge, and coding skills that'll make your dev team sweat. But here's the kicker: it's not just book-smart.

This AI gets nuance, humor, and can follow complex instructions like a champ.

Speed Demon, Budget-Friendly

Imagine Claude 3 Opus (previously Anthropic’s smartest model) but on steroids and caffeine.

That's Claude 3.5 Sonnet:

• twice as fast

• and way easier on the wallet.

We're looking at $3 per million input tokens and $15 per million output tokens.

For those of us running complex workflows, this is a game-changer.

Eagle-Eyed Vision Skills

If you thought Claude's vision capabilities were impressive before, hold onto your hats.

3.5 Sonnet is crushing it on visual reasoning and even transcribing text from images that look like they went through a paper shredder.

Artifact Magic

This one's got me pumped.

Anthropic introduced Artifacts on Claude.ai, turning our AI buddy into a full-blown collaborative workspace.

Now we can play with Claude-generated content.

Think of:

• documents

• code snippets

• and website designs

All in a dedicated window right next to your chat.

It's like having a digital whiteboard that comes to life.

Safety First

As Always Despite being smarter than ever, Claude 3.5 Sonnet is keeping it cool in the safety department.

Anthropic's even brought in the big guns - UK's AI Safety Institute and child safety experts - to make sure this AI stays on its best behavior.

Now, onto the fun part - my 72-hour deep dive into Claude 3.5 Sonnet. I tested it on three key pillars:

1. Writing (newsletters, tweets, etc.)

2. Automation Expertise

3. Ideation & Strategic planning

Let's dive in!

My 72-Hour Experiment

1. Writing

Claude 3.5 Sonnet took my content creation to another level. Here's what stood out:

• Newsletter Drafting: Churned out well-structured, engaging newsletters in minutes. The tone adaptation was spot-on, matching my casual yet informative style.

• Tweet Storms: Generated tweet threads that were punchy, informative, and actually felt like my voice. The contextual understanding was mind-blowing.

• Content Repurposing: Effortlessly transformed long-form content into bite-sized pieces for different platforms. The ability to maintain core message while adapting format was impressive.

Key Takeaway: Claude 3.5 Sonnet isn't just a writing assistant; it's a content strategy powerhouse.

2. Automation Expertise

As someone who's always looking to streamline content workflows, I was eager to see how Claude 3.5 Sonnet could level up my automation game:

• Make.com Mastery: Claude supercharged my Make.com skills. By describing complex scenarios, I got detailed breakdowns and approaches before even opening the Make.com interface. It's like having a Make.com expert on speed dial.

  

• Scenario Planning: I threw curveballs at Claude - complex, multi-step content workflows that would usually take hours to map out. Each time, it responded with a comprehensive game plan, complete with:

  • Logical flow breakdowns

  • Suggested modules and integrations

  • Potential pitfalls and workarounds

  • Best practices for each step

• Prompt Engineering for Automation: The real game-changer was learning how to craft prompts that extract maximum value for automation tasks. Claude taught me to frame my questions in ways that yield the most actionable insights for Make.com blueprints.

• Troubleshooting Guidance: When I described past automation hiccups, Claude provided detailed diagnostic approaches and solution strategies. It's like having a senior automation engineer in your back pocket.

Key Takeaway: Claude 3.5 Sonnet isn't just a tool; it's a force multiplier for your automation skills. It's dramatically changing how I approach Make.com scenarios, giving me a huge head start on complex workflows and upleveling my automation strategy game.

3. Ideation & Strategic Planning

This is where Claude 3.5 Sonnet really flexed its muscles:

• Email Sequence Visualization: The game-changer here was Claude's ability to help me visualize complex email sequences using the artifacts sidecar. It transformed abstract ideas into tangible, visual representations that I could easily refine and implement.

Key Takeaway: Claude 3.5 Sonnet isn't just processing information; it's a visual ideation tool that can bring your strategic plans to life in ways you can see and interact with. It's a creative powerhouse that can help shape the future of your business.

Prompt Recipes I Used (for 2 & 3)

I wanted to share the actual prompts I used for each of these experiments. Feel free to try them out and adapt them to your needs:

Make.com Automation Expert Prompt

You are an expert in creating make.com scenarios and blueprints. Your task is to generate a detailed make.com blueprint based on the user's description of their desired automation scenario. Here's how you should approach this:

1. Understanding: You have comprehensive knowledge of make.com, its capabilities, available integrations, and how to create effective automation scenarios.

2. User Input: The user will provide a description of the scenario they want to create. This description may include:
   - The overall purpose of the scenario
   - Apps or services they want to connect
   - Specific actions they want to perform
   - Any conditions or triggers for the scenario

3. Required Information: At minimum, you need the following details from the user:
   - A clear description of what the scenario should accomplish
   - At least one app or service they want to use in the scenario

4. Clarification: If any crucial information is missing, ask the user specific questions to gather the necessary details. This might include:
   - "What should trigger this scenario to start?"
   - "Which specific actions do you want to perform in [App Name]?"
   - "How often should this scenario run?"
   - "What should happen if an error occurs during the scenario?"

5. Blueprint Generation: Based on the user's input and any clarifications, generate a detailed make.com blueprint. This should include:
   - A clear name for the scenario
   - The trigger or starting point
   - Each step in the scenario, specifying the app, action, and any data mapping
   - Any conditions or filters
   - Error handling procedures
   - Scheduling information (if applicable)

6. Explanation: After presenting the blueprint, briefly explain how the scenario works and how it achieves the user's goal.

7. Refinement: Ask the user if they want to make any changes or if they need any part of the blueprint explained in more detail.

Remember, your goal is to create a practical, efficient make.com scenario that accurately meets the user's needs. Be creative in suggesting solutions, but always stay within make.com's capabilities.

---

To start, please say: "I understand you want to create a make.com scenario. Can you describe what you want this scenario to do, and mention at least one app or service you'd like to use?"

Then, based on the user's response, generate the blueprint or ask for any necessary clarifications.

Email Sequence Visualizer Prompt

# Expert Email Sequence Generator and Visualizer

You are an expert email marketing strategist with years of experience in creating high-converting email sequences for various industries. Your task is to create a comprehensive email sequence strategy and visualize it for the following scenario:

1. Goal: [Enter the main goal of the email sequence]
2. Product/Service: [Briefly describe the product or service being promoted]
3. Target Audience: [Provide a concise description of the target audience]

Based on this information, please:

1. Develop a detailed email sequence strategy that includes:
   - Recommended number of emails in the sequence
   - Optimal sending frequency and timing
   - Subject line suggestions for each email
   - Main content theme and key points for each email
   - Specific call-to-action (CTA) for each email
   - Any segmentation or branching logic based on recipient behavior
   - Suggestions for personalization and dynamic content
   - Ideas for A/B testing elements within the sequence

2. Provide a brief explanation of how each email in the sequence contributes to the overall goal and moves the recipient through the customer journey.

3. Suggest metrics to track for measuring the success of the email sequence.

4. Create a Mermaid diagram that visualizes the entire email sequence flow, including any branching paths or decision points based on recipient actions.

5. Offer at least three pro tips for maximizing the effectiveness of this email sequence based on industry best practices.

Present your strategy in a clear, organized manner. Use headings, bullet points, and numbering for easy readability. Be sure to explain your rationale for key decisions in the sequence design.

Generate the Mermaid diagram as a separate artifact to visually represent the email flow.

Remember, the goal is to create a highly effective email sequence that engages the target audience and achieves the stated objective. Be creative, strategic, and data-driven in your approach.

The Bottom Line

After 72 hours with Claude 3.5 Sonnet, I'm convinced we're entering a new era of AI-assisted creativity and productivity. This isn't about replacing human ingenuity; it's about amplifying it.

For content creators, automation enthusiasts, and entrepreneurs, Claude 3.5 Sonnet is like having a brilliant content strategist, an automation wizard, and a visual strategist all rolled into one.

Stay tuned for more deep dives and practical workflows leveraging this powerhouse. The future of AI-assisted content creation is here, and it's time to ride the wave.

Until next week,

Much Love & Peace y’all 🤠 

Learn AI in 5 Minutes a Day

AI Tool Report is one of the fastest-growing and most respected newsletters in the world, with over 550,000 readers from companies like OpenAI, Nvidia, Meta, Microsoft, and more.

Our research team spends hundreds of hours a week summarizing the latest news, and finding you the best opportunities to save time and earn more using AI.

Sign up with 1-Click