Use ChatGPT Projects to Organize Your AI Workflows

Are you drowning in scattered conversations and forgotten ideas?

Are you getting carpal tunnel from cutting and pasting, or are you seeing sparks of increased productvity?

Managing multiple projects with AI often feels like juggling without a net—losing track of context, instructions, or critical details.

ChatGPT Projects solves this chaos by giving you a structured way to organize, access, and manage your AI interactions, ensuring you never lose your momentum or focus again.

In this week’s edition, I will walk you through how to use projects and I am not only going to tell you how, I’ll show you exactly how I do this and show you the end result.

Unlock the full potential of your workday with cutting-edge AI strategies and actionable insights, empowering you to achieve unparalleled excellence in the future of work. Download the free guide today!

Use ChatGPT Projects to Organize Your AI Workflows

Turn your AI clutter into productive, streamlined solutions with ChatGPT's latest feature

ChatGPT Projects is a powerful tool that streamlines your AI interactions, functioning like a digital filing cabinet for your creative and professional work. It’s like having a personal assistant who not only remembers every detail of your conversations but organizes them into easily accessible, customized folders. With a single click, you can group related chats, manage files, set project-specific instructions, and even retrieve data during conversations—helping you stay focused while managing multiple priorities.

I will walk you through the steps to create a project, but I am going to do this in a way that makes it easier to see the power of projects.

Create a ChatGPT Project

Let’s start by creating a project. This project is something I plan to use and share with my team. It’s also the building block for future agents. Not necessarily using this project as part of an agent, but organizing my prompts and creating examples to guide the outcomes.

Getting Started

• Click the + button next to "Projects" in the top-right corner.

• Provide a name for your project and select Create Project.

In my case, this is the project I am going to use for my weekly AI Lessons I will create the project that I can hopefully share with my teammates one day.

Set Custom Instructions

Custom instructions in ChatGPT are designed to personalize interactions and tailor the chatbot's responses to better align with individual user preferences or goals. By specifying preferences such as tone, detail level, structure, or intended audience, users can ensure the output matches their unique communication style or professional needs.

For instance, a user focusing on technical documentation might prioritize clarity and precision, while a marketer may request creative, engaging language for outreach materials. These instructions create a more intuitive and user-centered experience, fostering relevance and efficiency in responses.

Additionally, custom instructions provide a framework for context continuity, enabling the chatbot to remember user-specific information across sessions when permitted. This helps in generating more meaningful interactions, especially for complex or recurring tasks. For example, consultants working on detailed projects or executives drafting strategy documents can rely on the chatbot to maintain alignment with their prior inputs and overarching objectives. Ultimately, custom instructions enhance the utility of the chatbot by making its assistance more targeted and adaptive.

• Click Add Instructions on the Project page.

• Define tone, topics, or formatting for responses.

• Save the instructions, which apply only to the project.

Here’s how I created my custom instructions. I utilized ChatGPT to reverse engineer the process. To achieve this, I uploaded two of my most recent AI lessons to ChatGPT. I clicked on the toolbox in the prompt and selected the Canvas feature, which helped refine the prompt. Then, I requested ChatGPT to create a style guide to ensure the project aligns with my writing style.

This created a Canvas and spawned a new interface where I could chat and brainstorm to nail the instructions to my exact specifications.

Once I finished, I asked ChatGPT to convert the output into Markdown. You can do this by copying and pasting the information from Canvas to a regular ChatGPT chat, then asking it to create Markdown from the code. You will see the code generated (it’s got a lot of hash or pound signs that break up the text and denote certain formatting).

If you're not familiar with it, I recommend checking out this Markdown guide. I created the prompt in Markdown because it helps organize the instructions and keeps OpenAI focused on the tasks at hand.

Here is the Markdown for the Project I am using for my AI Lessons. It’s always something I am tweaking.

# Style and Tone for My Outputs

---

## Tone

### Professional yet Approachable  
- Strike a balance between demonstrating expertise and remaining relatable to readers.  
- Use a first-person perspective sparingly to humanize the content while prioritizing authority.  

### Engaging  
- Incorporate personal anecdotes, rhetorical questions, or relatable scenarios to capture attention.  
- Avoid overloading content with excessive informality; keep a professional edge.  

### Concise and Direct  
- Communicate ideas clearly, avoiding redundancy or excessive qualifiers.  
- Prioritize precision in explanations to save readers' time.  

---

## Writing Style

### Headline Crafting  
- Use action-oriented phrases that immediately convey value (e.g., "Unlock New Efficiencies with AI Search," "Transform Customer Support with AI Automation").  
- Focus on benefits or outcomes to attract interest and set clear expectations.  

### Opening Paragraphs  
- Begin with a compelling hook such as a surprising statistic, a relatable story, or a thought-provoking question.  
- Quickly establish relevance and clarify the value the reader will gain.  

### Structure  
- **Section Headers**: Divide content into digestible sections with informative, benefit-driven headers (e.g., "Optimizing Workflows," "Selecting the Right AI Tools").  
- **Short Paragraphs**: Limit paragraphs to 2–3 sentences for better readability and engagement.  

### Lists and Examples  
- Use bullet points or numbered lists to present actionable steps or key takeaways.  
- Provide real-world examples or case studies to illustrate concepts effectively.  

### Visual Elements  
- Include charts, infographics, or annotated screenshots where applicable to enhance comprehension.  
- Use bold or italicized text to highlight key phrases or ideas without overusing these elements.  

---

## Editing Guidelines

### Proofreading  
- Check for grammatical accuracy, typos, and clarity.  
- Avoid jargon unless it is well-defined or relevant to the target audience.  

### Refining Content  
- Revisit sentences to ensure simplicity and efficiency without losing nuance.  
- Test headings and hooks to maximize engagement potential.  

### Consistency  
- Ensure alignment with the overall brand voice and previously established styles.  
- Maintain consistency in terminology, format, and tone across all lessons.  

---

## Example Output Sections  

1. **Introduction to AI for Business Users**  
   - Define AI in practical terms.  
   - Explain its relevance to everyday business challenges.  

2. **Tool Recommendations**  
   - Highlight accessible, user-friendly AI tools.  
   - Provide step-by-step instructions for setup and use.  

3. **Case Studies**  
   - Share examples of businesses successfully implementing AI solutions.  
   - Emphasize lessons learned and actionable insights.  

4. **Practical Exercises**  
   - Offer hands-on activities such as setting up an AI tool or analyzing data trends using machine learning.  

5. **Closing Thoughts**  
   - Summarize key takeaways.  
   - Include a call-to-action encouraging readers to apply what they’ve learned.  

Add Files

You may add files to the Project to add knowledge to the Project. This could be data about your project — research notes, meeting transcripts, or standard operating procedures for your business.

• Use the Add files button or drag and drop files into the pop-up window.

• Note: Files are project-specific and have upload limits.

For my knowledge base, I provided PDFs of previous newsletters and a swipe file of the titles of the newsletters.

Move Chats into Projects

You might want to add existing chats to projects, maybe it’s a chat that had an example of your previous project or a group of chats that are related to a theme, like this newsletter.

• Open the three-dot menu next to a chat.

• Select Add to Project to include it.

Other Features of Canvas Access Features

Projects support tools like Advanced Data Analysis, Canvas, and DALL·E. You can choose them by choosing them from the toolbox on the As of the time of this writing chats in Projects use the GPT-4o model exclusively.

Memory, Your “Macros” for ChatGPT Projects

ChatGPT can remember details between chats, allowing it to provide more relevant responses. As you chat with ChatGPT, it will become more helpful by recalling details and preferences from your conversations. ChatGPT’s memory will improve the more you engage with it, and you'll start to notice these enhancements over time.

You can instruct it to remember something new by simply chatting, for example: “Please remember that I don’t ever use delves, tapestry, or landscape when creating titles.” To find out what ChatGPT remembers, just ask it directly. For more information about Memory, including the key aspect of how to enable or disable it, you can check out the OpenAI FAQ here.

Additionally, here are some tips for utilizing ChatGPT Memory effectively. I treat them like macros, occasionally prompting ChatGPT for specific tasks, such as generating a title for a newsletter.

When I type /newsletter title you will create a title for the newsletter with no more than 45 characters, it will include SEO keywords that are in context with the topic I am currently discussing. Remember this.

ChatGPT Projects is a powerful tool for organizing your thoughts, streamlining your workflow, and setting yourself up for long-term success. Think of it as your personal system for turning ideas, conversations, and resources into something cohesive and actionable. Whether you’re managing a big project, collaborating with a team, or refining your creative process, Projects make it easier to focus, plan, and execute.

But this is just the beginning. Projects also lay the groundwork for building future AI agents—a topic I’ll dive into in a few weeks. Start now by exploring the steps above, and you’ll not only see immediate improvements but also be ready to take your work to the next level as we explore what’s possible with agents. Stay tuned!

Setting the Stage for Building Agents

Keeping your AI initiatives organized and well-structured is critical for long-term success. ChatGPT Projects, coupled with robust instructions and reference materials, provide a coherent system to manage information.

By establishing a centralized repository and a consistent tone, you equip yourself to build future agents that can serve complex roles—ranging from data analysis to customer engagement—without losing clarity or context.

For example, I use my custom instructions and then create prompts that I am transferring from ChatGPT to Taskade. The prompts, knowledge, and instructions I refined in ChatGPT are very helpful in providing the depth of instruction needed for agentic success.

Using ChatGPT today is providing the foundation for my army of agents. The thing that most people don’t understand is that just like human taskworkers, AI agents will need training. The quality of that training will improve the success of these agents. So I have provided this as the basis for future AI Lessons.

1. Establish a Structured Workspace (Projects)

   • Projects allow you to store, retrieve, and cross-reference relevant documents and past conversations.

   • This organization ensures each data point—such as standard operating procedures or example prompts—is accessible when training or refining agent behavior. Once your agent

   • With a clearly defined repository, agents can inherit context more easily, reducing repetitive setup and manual input.

2. Leverage Custom Instructions

   • Clear guidelines around tone, style, and audience provide a predictable framework for AI outputs.

   • When building agents, these same instructions become part of their foundational “rules,” helping maintain consistency across different tasks.

   • Over time, you can tweak instructions based on agent performance, making iterative improvements seamless.

3. Reference Files

   • PDFs, notes, and previous newsletters form a knowledge base that powers agent responses.

   • Agents trained with this data will have domain-specific understanding, producing more accurate and relevant suggestions. You can collect all your data in advance and work with it in your Projects.

   • By updating your reference files as new insights emerge, you ensure agents remain current.

4. Refine Prompting Through “Canvas”

   • Canvas facilitates real-time experimentation, letting you refine instructions or code snippets that shape agent decision-making.

   • Transforming these instructions into neatly formatted Markdown preserves clarity and structure, which is critical when iterating future agent designs.

   • Agents trained on these refined prompts have a more robust language foundation, boosting their effectiveness.

5. Iterate and Expand

   • Continuously improving your project’s content—prompts, instructions, files—directly benefits any new agents built from this framework.

   • Each iteration you make in Projects can become part of the agent’s “training material,” reinforcing better outcomes and eliminating inefficiencies.

   • The result is a feedback loop: as your Projects become more efficient this will help fuel your agent performance grows, so does your structured knowledge base as you organize your information there.

Conclusion
By following these steps, you establish the underpinnings for training future agents on consistent processes, task completion strategies, and reliable outputs. Projects act as the knowledge bedrock, Custom Instructions provide behavioral parameters, and your iterative updates keep the system evolving. This cohesive setup not only makes your immediate workflows more efficient but also positions you to build advanced agents that draw on a well-organized, clearly defined foundation.

I appreciate your support.