Creating AI Assistants Using Claude Projects

Want to create consistently high-quality content while maintaining your unique tone and brand identity? Are you looking for ways to streamline your workflow with AI assistants?

In this article, you'll discover how to create AI assistants using Claude Projects.

Claude Projects: A Game-Changer in AI Assistance

Since the release of ChatGPT in November 2022, the journey of AI in business has been a rollercoaster of expectations and realities. However, the initial excitement was often followed by disappointment because AIs couldn't consistently produce output that could be used in a workflow.

This disillusionment led many businesses to hesitate to embrace AI tools fully. However, recent advancements, particularly with tools like Claude Projects, are changing our reliance on AI-generated content for business operations.

Claude, developed by Anthropic, is a powerful AI model that competes with OpenAI's ChatGPT. 

While both offer AI assistance, Claude Projects provides unique advantages for businesses.

First, Claude can process and remember more information within a conversation, leading to more coherent and contextually relevant outputs.

Then, there's the writing quality. Many users find Claude's writing more natural and human-like than ChatGPT's. In particular, Claude excels at mimicking specific writing styles, making it ideal for brand-consistent content creation.

Additionally, unlike ChatGPT, Claude Projects can't search the web. This perceived limitation is actually a benefit, as it ensures that the AI works only with the specific, high-quality data you provide instead of adding irrelevant or potentially off-brand content to your writing.

#1: Step-by-step guide to Getting Started with Claude Projects:

Note: You'll need a paid Claude account (currently $20/month for a professional account) to access Claude Projects.

To create a powerful AI assistant using Claude Projects, you'll start by creating the Project Knowledge base. Here, you will create custom instructions and document uploads that provide brand details and examples Claude can learn from.

Kyle Balmer recommends following the SPUN framework for this process:

S – Scope: Define your AI assistant's specific task or role. P – Priming: Provide detailed instructions and context to guide the AI's behavior. U – Uploading: Add relevant content to your Claude Project. N – Narrowing: Focus your assistant on specific areas of expertise.

Let's break down each step in detail:

Scope

Scope is about determining what you want the assistant to do and working within the AI's limitations. Since Claude cannot connect to the internet, remove any task that requires an internet connection from the possibilities.

When defining the scope, start by identifying repetitive, time-consuming tasks in your workflow. These are often the best candidates for AI automation. Look for tasks that take up much of your time but differ from your favorite parts of content creation, like improving writing for landing pages or captions for social media posts.

As you decide what type of assistant to build, keep these tips in mind:

• Be specific about the task (e.g., “Email Marketing Specialist for Product X” rather than just “Marketing Assistant”)
• Consider the limitations of the AI (e.g., no web searching capability)
• Think about the end goal for this assistant. 

Examples might include email marketing specialist, social media content creator, product-specific marketing assistant, newsletter writer, or customer FAQ responder.

Priming

Priming tells the AI what you want it to do and how to do it. Balmer uses the RISEN framework for effective priming:

• Role: Define the specific job of the AI assistant.
• Instructions: Provide detailed guidance on how to approach the task. Specify any constraints or limitations (e.g., word count, tone, formatting)
• Steps: Break down the instructions into clear, sequential steps.
• End goal: Clarify what success looks like for the AI.
• Narrowing: Focus the AI on specific aspects of the task.

Add these details to the project's custom instructions.

Your role is to act as an expert newsletter writer for a tech startup.

I want you to write engaging, informative newsletters about AI advancements in business.

Follow these steps:
Start with an attention-grabbing headline
Provide a summary of the main topic
Dive into 2-3 key points with examples
Include a call to action at the end

Your end goal is to create newsletters that increase open rates by 20% and click-through rates by 15%.

Focus on AI applications in marketing and customer service.

Uploading

Now, you're ready to upload examples of quality work you want the AI to imitate.

The more information you give the AI, the more it has to work with. That said, the data you upload must be very relevant and high-quality content relevant to the task.

If you're building a newsletter writing assistant, you would include past newsletters, brand guidelines, product information, a customer persona, etc.

Steps for effective uploading:

1. Gather relevant documents (e.g., past newsletters, brand guidelines, product information)
2. Clean the data (remove irrelevant information, formatting, etc.). If you are uploading information from a website, strip out all the HTML and CSS and any metadata.
3. If necessary, break large documents into smaller chunks (Claude has a 10MB per file limit).
4. Upload the documents to your Claude Project.
5. Verify that the uploads were successful.

Narrowing

It's essential that you build an assistant that does one task well. The more precise you are, the better the quality of the AI's output. That's where narrowing comes in. 

Test the output and edit the custom instructions to make any necessary adjustments.

Provide Iterative Feedback

To refine your AI assistant's output, treat it like a new team member: provide feedback on its output and ask for revisions to help it learn your preferences.

Balmer explains his process:

“I will copy the final version back into my Claude assistant and say, ‘Hey, this is what I landed on, so this includes my human edits; learn from this.' And it will go through and compare to the last version with the final published version, and it will summarize, ‘Okay, these are the changes you've made; I will include these in the future.'”

Follow this step-by-step feedback process to get high-quality results:

1. Review the AI's initial output.
2. Identify areas for improvement.
3. Provide specific feedback (e.g., “Make the intro more casual” or “Add more data to support point 2”)
4. Ask for a revised version.
5. Repeat steps 1-4 until satisfied.
6. Share the final, human-edited version with the AI for learning.

Pro Tip: Don't be afraid to go back and forth with your AI assistant multiple times. The more feedback you provide, the better it will learn your preferences and style. Kyle Balmer shares that he sometimes goes through twenty or thirty rounds of feedback to write his newsletter. Some changes will be small, such as:

You missed the introduction; please add an introduction," and some will be larger, such as "I want you to shuffle [these sections].

Use Natural Language

When giving instructions or feedback, use conversational language. Claude is designed to understand and respond to human-like communication, so even if you don't necessarily know what's off with the content, you can say, ‘I don't like that. There's something wrong with the first paragraph.'”

Other examples of natural language feedback:

• “This intro feels too formal. Can you make it more conversational?”
• “I like the main points, but can you add more specific examples?”
• “The conclusion seems abrupt. Can you smooth it out and add a stronger call to action?”

#2: Practical Implementation of Claude Projects in Your Workflow

Once you've setup your Claude Project, here are some detailed tips for integrating it into your workflow.

Maintaining Brand Voice and Quality

While Claude is excellent at mimicking styles, ensuring consistent brand voice and quality across all outputs can be challenging.

Tips for maintaining brand voice:

• Provide clear examples of your brand voice in the project data
• Use the iterative feedback process to refine the AI's understanding of your style
• Create a “style guide” document specifically for your AI assistant
• Regularly review outputs to ensure consistency

Leverage the Mobile App

Claude's mobile app allows you to work on projects on the go, making it easy to manage your AI assistants from anywhere.

“I actually do most of my newsletter writing on the mobile app, which sounds absolutely wild, but it's such a nice interface that it's entirely possible to work with projects on a telephone,” Balmer shares. “So just having it there on my phone and being able to pick up a chat that I've been having on the computer anywhere and kind of go through its latest draft and say, maybe you can change this. Maybe, you know, you can change that.”

Tips for using the mobile app:

• Use voice dictation for quick input.
• Review and provide feedback during commutes or downtime.
• Save drafts for later refinement on your computer.

Combine Claude Project With Other Tools

While Claude Projects are powerful, they can be even more effective when combined with other AI tools.

For example, use Claude to generate the initial content and paste Claude's output into ChatGPT for refinement. Then, take ChatGPT's refinements back to Claude for analysis and learning. 

After you've made your final edits manually, paste the final version into Claude for future improvement.

Please compare your version with this final version. What differences do you see? Use this final version to improve your writing.

Potential AI tool combinations:

• Claude for initial content creation + ChatGPT for editing and refinement
• Claude for long-form content + Midjourney for image generation
• Claude for data analysis + Tableau for visualization

Share Projects With Your Team

Claude Projects can be shared within your organization, allowing for collaborative AI-assisted work. These steps will let you collaborate using Claude Projects:

1. Create a shared project for your team.
2. Invite team members to the project.
3. Establish guidelines for how to interact with the AI assistant.
4. Use the shared project for collaborative tasks (e.g., content creation, brainstorming).
5. Regularly review and refine the project based on team feedback.

#3: Overcoming Limitations and Maximizing Claude's Potential

While Claude Projects offer numerous advantages, it's important to be aware of their limitations and how to work around them.

The Learning Curve

Creating effective AI assistants requires practice and experimentation. Don't be discouraged if your first attempts don't yield perfect results.

Strategies for mastering Claude Projects:

• Start with simple, well-defined tasks and gradually increase the complexity.
• Experiment with different priming techniques to see what works best.
• Keep a log of successful prompts and instructions for future reference.
• Regularly review and refine your projects based on their performance.

No Web Search Capability

As mentioned previously, unlike ChatGPT, Claude cannot search the internet for information. This means you need to provide all necessary data upfront.

However, this limitation can also be advantageous, ensuring that the AI only works with the specific, high-quality data you provide.

Strategies to overcome this limitation:

• Regularly update your project with new, relevant information.
• Use the built-in screenshot tool to capture and upload web content directly into your project.
• Combine Claude with other tools that can search the web when needed.

File Size Limitations

There's a 10MB limit per file upload. Tips for managing file size:

• Break large documents into smaller chunks
• Remove unnecessary formatting and images before uploading
• Use text-based formats (e.g., .txt, .md) instead of rich formats like .docx
• Compress images before including them in your uploads

Pro Tip: To overcome the file size limitation and make the most of your Claude Project, consider creating a “knowledge base” document summarizing critical information from multiple sources. This can serve as a compact yet comprehensive resource for your AI assistant.

Other Notes From This Episode

• Connect with Michael Stelzner @Stelzner on Facebook and @Mike_Stelzner on X.
• Watch this interview and other exclusive content from Social Media Examiner on YouTube.

Where to subscribe: Apple Podcasts | Spotify | YouTube Music | YouTube | Amazon Music | RSS

✋🏽 If you enjoyed this episode of the AI Explored podcast, please head over to Apple Podcasts, leave a rating, write a review, and subscribe.

---

Stay Up-to-Date: Get New Marketing Articles Delivered to You!

Don't miss out on upcoming social media marketing insights and strategies! Sign up to receive notifications when we publish new articles on Social Media Examiner. Our expertly crafted content will help you stay ahead of the curve and drive results for your business. Click the link below to sign up now and receive our annual report!