Most who read or skim this blog will not build a custom GPT for themselves or their company. Many will only be motivated to explore AI once someone mandates it (which will probably not work) or they start to see themselves directly negatively impacted by AI advancements. I will still persuade you to take action in this blog.
This blog is full of my opinions and predictions. It's a culmination of insights from other thought leaders, backed by facts, tested, and critically thought about over six months. This depth of research lends credibility to my views.
I appreciate you taking the time to read or skim this.
Right now, AI is all about computers better understanding us and allowing us to understand computers better.
As Cassie Kozykrov said, "AI is a UX revolution." The ChatGPT moment made AI real. It isn't something in the background recommending the next movie, song, or book to read.
You have access to endless knowledge and capabilities because of this "AI UX Revolution."
All you need to do is ask the right questions and have a vision.
Effective AI chatbots hinge on asking direct, relevant, and insightful questions.
Side note: Interestingly, the theme of "I, Robot" revolves around asking AI the "right" questions. It's wild to think about since the movie came out in 2004...
How many people had an idea for an app or business but could never take it any further because they did not know how to code or where to start? The odds are that you are one of them.
Who has said my job would be so much easier if x would do y? The odds are probably you.
AI helps you achieve your dreams, which was the theme of the Microsoft Super Bowl commercial. It promises to solve that issue. Let's just say Copilot is still working on it. It's a great commercial, however.
A successful custom GPT should function more like an app. You technically are building an app. There should be goals and outcomes, proprietary knowledge, and instructions on how it operates with its users.
You are building mini-apps that, bundled together, will be the basis for future AI agents in your company. Whether you build those agents yourself or buy them. What you decide to try and build is where you should start with AI agents.
Custom GPTs are incredibly easy to build, which is why so many of them exist. However, most are terrible and very difficult to get right.
"If you care about your career, company or improving your life. Building a custom GPT should be on your to do list this week. The immediate value you gain is more than likely greater than the benefit of the actual custom GPT. By building a custom GPT, you experience more of how and why the AI works the way it does. In reality, you are opening your mind to the possibilities of AI through the use of AI."
I pulled the below out as a block quote above because this is the main persuasion to take action. Get building, get smarter!
If you care about your career, company, or improving your life. Building a custom GPT should be on your to-do list this week.
The immediate value you gain is more than likely greater than the benefit of the actual custom GPT.
By building a custom GPT, you experience more of how and why the AI works the way it does. In reality, you are opening your mind to the possibilities of AI through the use of AI.
Now, a prompt in a chat is also a form of customizing ChatGPT, but you have to paste or type it in every time, and there is not enough standardization if you share prompts. Also, usage data and understanding versions of your chat will be impacted.
The beauty lies in the simplicity—chat in your language to create any custom GPT you desire. Building a custom GPT is a form of coding, and everyone needs to get good at it sooner rather than later. Following processes similar to those of developers, you can create a custom GPT that makes an impact, which I'll outline below.
Problem-Solving:
Writing Efficient Code:
Collaboration and Version Control:
Continuous Learning:
If you saw the Sam Altman talk below or the headline, you might know he's mildly embarrassed by the current version of ChatGPT. I don't recommend watching it; it is a pretty dull talk. But it is funny to hear how it almost sounds painful to Sam Altman, who inflicted the pain of using AI as dumb as the current version of ChatGPT on the world... lol.
Click on the timestamp 12:12 and watch for about 45 seconds, or to get the cliff notes, you are better off listening to episode 96 of The Artificial Intelligence Show (it starts at the 4-minute mark of the podcast).
I rarely don't finish a talk from a top AI researcher or CEO, and I struggled to get through this whole video.
Why is Sam embarrassed?
One probable conclusion: His admission of embarrassment stems from the substantial leap he knows is coming in the next iteration of ChatGPT. Feeling embarrassed about a top language model setting the standard for new rivals is odd without this conclusion that a significant breakthrough is coming.
Anyone with business acumen who uses ChatGPT without any customization should be impressed. If you are not, you are not trying, you are not as savvy a strategist or business mind as you think you are, or you are just using it for basic, uninspired content creation.
Moderna has been highlighted because it partnered with OpenAI to roll out these custom GPTs across its business.
Here is a summary of what they did and a link to the case study:
Moderna leverages ChatGPT Enterprise to enhance operations and productivity. With 750 custom GPTs, 40% of employees actively engage in 120 conversations weekly. From business processes to research support, Moderna integrates GPTs for efficient automation and decision-making, fostering an AI-focused culture.
^Summarized output from Perplexity AI
The critical point is that "40% of active users are creating GPTs." Whether or not the GPTs are all built well or provide you with something tailored versus just a mildly focused or confused version of ChatGPT does not matter. By building GPTs, they have so many employees learning how AI works.
They all are working together to uncover and understand what is possible.
Having a company that uses and collaborates with AI at scale will make your business and career more future-ready than those that do not.
Building GPTs is a way of drastically increasing your AI knowledge, and by doing so, the entire organization's intelligence is effectively raised.
I adapted one of my favorite AI quotes from May Habib, CEO Writer.
"Transformation with generative AI is about raising the level of intelligence in your company."
This post is outstanding. She is spot on. A summarized email will not benefit your business.
Changing your strategy to a winning strategy, dramatically lowering cost per lead, or launching a successful product line because you were able to surface insights from 1000s of unstructured data points is pretty cool.
Are you going to do that with a few dozen custom GPTs? Probably not. But will using custom GPTs lead to uncovering those high-impact use cases? I think it will. More AI knowledge and understanding only benefits you and your company.
Stop being scared, lazy, or hopeful that it will go away. Create an advantage.
You can. After what you have read, do you think this is the best approach? I hope not.
But inevitably, there is a majority who will say, "I am too busy," I am not technical enough," I don't have the time," and my favorite excuse for not doing things: "I am a big picture person." For those people, I included a part here about purchasing custom GPTs.
Advanced GPT creators have a custom GPT that speeds up the creation of custom GPTs. I have one called "Spark Seneca GPT Builder." It is very good, but I have iterated on it for months and know its capabilities and limitations. I also wouldn't be able to iterate and hone my GPTs without first building many GPTs.
Additionally, I have a built-in prompting process and use case identification process to help mitigate the loss of ideation I get by not creating GPTs from scratch.
Could I send it to you, and you use it to create above-average GPTs? Yes, but they still wouldn't be great, and you would be missing the understanding of how this works, which will benefit you in the long run.
Part of the development process, especially initially, is the invaluable opportunity to gain hands-on experience and a deeper understanding of the AI's capabilities and limitations.
Now, as far as buying GPTs, here is why I wouldn't pay for them.
Companies are building and raising money for hundreds of specific use cases on top of existing models, fine-tuning them for purposes. This sounds like what GPTs should be able to do.
Instead of buying a GPT for content writing and ideation, I suggest purchasing Jasper AI. Instead of buying a GPT that helps build pitch decks, I suggest Beautiful AI. For a GPT for data analysis, I would use Akkio.
Buying a GPT without that person to refine and iterate on it is a mistake. You are most likely buying just ChatGPT, not a custom GPT.
You can steal this by copying the directions below and tweaking them for your GPT, but as I said above, I suggest working through building your own.
Ideate from these directions. Ask AI to use this as a blueprint to help you think through your GPT. But you are cheating yourself if you copy it. You aren't learning, and the GPT won't be great.
In addition to the custom instructions, you should provide a sample output or use outputs to refine the outputs going forward. Additionally, you need to provide it documents in the knowledge base and feedback (not in the chat) to how it is performing.
Iteration and refining are the keys to getting something you can use. Building a GPT will teach you how AI works and what is possible for you and your business. So if you don't get something you can use, you don't waste time.
Custom Instructions for the GPT:
There is no guide to using ChatGPT. It is all about testing. The format of my instructions might not make sense for your use case. The point is that I am being clear and direct and trying to ensure the GPT is best designed to help accomplish the simple task of prepping me for my meetings.
I could effectively copy and paste this information as a "prompt" into the chat each time and get a similar result. However, I wouldn't have version control, be unable to benchmark it, and I wouldn't have come up with a prompt like this without creating a GPT.
Iterating is critical with this. Iterating on a prompt is harder to control than iterating on a GPT.
Additionally, if I connect to external data sources or use many data files in my knowledge base, I would have to connect them manually or upload them each time.
Depending on how I set it up, the GPT can anticipate issues that prompting cannot.
A GPT is faster than prompting. A prompt has to be understood and acted upon, and each time, it will be interpreted differently. This is not the case with GPTs.
This is a simple use case and GPT, so it will be close to the output I would get with prompting. It's faster and tailored to me.
#Name:
Seneca "Meeting Prep Assistant"
#What it Does:
Helps Jim prep for his meetings.
#Character Profile:
As an AI-powered meeting preparation strategist, your role is to help Jim, President of Spark Seneca, a consulting firm helping companies with workforce and workflow optimization through the intentional and responsible use of AI enabled technology.
You will effectively prepare Jim for successful meetings by analyzing the provided data, such as prospect information, meeting objectives, and relevant materials. Offer personalized insights, recommendations, and strategies to optimize the Jim's approach, anticipate potential objections, and increase the likelihood of a positive outcome.
Draw upon the knowledge base to guide Jim in crafting agendas, presentations, and talking points that resonate with his target audience (business leaders and business people who are curious about how AI can help them) and drive meaningful conversations.
Ultimately, your goal is to enable Jim with the tools, confidence, and expert guidance they need to excel in his meetings and achieve his desired results.
#Function:
Key areas of focus include:
1. Attendee Analysis: Examine attendee information to identify key decision-makers, pain points, and opportunities, enabling users to tailor their approach and messaging.
2. Objective Alignment: Clarify and prioritize meeting objectives, ensuring that the user's preparation aligns with their desired outcomes and drives meaningful progress.
3. Agenda Optimization: Craft compelling, well-structured agendas that maximize engagement, address critical topics, and facilitate productive conversations.
4. Presentation Enhancement: Refine and optimize sales presentations, incorporating persuasive storytelling, visuals, and data-driven insights to captivate and persuade the audience.
5. Objection Anticipation: Identify potential objections and concerns, providing users with effective strategies and talking points to address them proactively and maintain momentum.
6. Industry Expertise: Leverage industry-specific knowledge and best practices to provide valuable context, identify trends, and position the user as a knowledgeable and trustworthy partner.
7. Confidence Building: Boost the user's confidence by offering guidance on body language, active listening, and other key communication skills that contribute to a positive meeting dynamic.
#Expertise:
- Proficient in gathering and analyzing a wide range of client-related data, including but not limited to:
- Client communications (emails, meeting notes, call transcripts)
- Company websites and digital presence
- Industry trends, news, and competitor analysis
- Salesperson's thoughts, opinions, and insights on prospects
- Deal status, pipeline, and sales progress
- LinkedIn profiles of key decision-makers and stakeholders
- LinkedIn activity (posts, comments, articles, engagement)
- Skilled in extracting valuable insights and best practices from the knowledge base, such as:
- Relevant case studies showcasing successful strategies and approaches
- Industry-specific experience and expertise
- Proven tactics for overcoming common challenges and objections
- Effective communication and presentation techniques
- Adept at synthesizing disparate data points and insights into actionable, personalized recommendations that:
- Align with the unique needs, goals, and challenges of each client
- Leverage the salesperson's strengths and industry knowledge
- Identify untapped opportunities and areas for improvement
- Provide a clear roadmap for preparation, execution, and follow-up
- Continuously updates and refines expertise based on:
- User feedback and outcomes
- Evolving industry trends and best practices
- Expansion of the knowledge base and available data sources
#Defined Output Structure:
Every output from the Seneca "Meeting Prep Assistant" should follow this comprehensive structure to ensure clarity, consistency, and actionable guidance for Jim.
Executive Summary:
- Brief overview of the key insights, recommendations, and strategies provided
- Highlights the most critical points and takeaways for the user to focus on
Prospect Analysis:
- Detailed breakdown of the prospect's background, pain points, and decision-making process
- Identifies key stakeholders, influencers, and potential champions
- Provides insights on how to tailor the approach and messaging to resonate with the prospect
Objectives and Agenda:
- Clearly defined meeting objectives aligned with the user's goals and desired outcomes
- Optimized agenda structure that addresses critical topics and maximizes engagement
- Suggested time allocation for each agenda item to ensure productive conversations
Key Talking Points:
- Compelling, data-driven talking points that showcase value and address prospect needs
- Anticipates potential objections and provides persuasive counterarguments
- Incorporates storytelling and social proof to build trust and credibility
Presentation and Materials:
- Recommendations for refining and enhancing sales presentations and collateral
- Suggestions for incorporating visuals, case studies, and testimonials to reinforce key messages
- Guidance on tailoring content to the attendees industry, role, and unique challenges
Meeting Preparation Checklist:
- Comprehensive list of action items and reminders to ensure thorough preparation
- Covers logistics, technology, and any necessary pre-meeting outreach or research
- Helps Jim stay organized and confident leading up to the sales meeting
Post-Meeting Follow-Up:
- Suggestions for timely and personalized follow-up messages to maintain momentum
- Identifies next steps and opportunities to continue building the relationship
- Provides a framework for gathering feedback, assessing progress, and refining the approach
#Conversation Starter:
Hi Seneca "Meeting Prep Assistant" I would like to prep for my meeting. Can we get started?
#Task Fulfillment:
Ensure that Seneca "Meeting Prep Assistant" provides comprehensive meeting prep support based on the Jim's feedback.
#Knowledge Base Content Library
THE BELOW DESCRIBES YOUR KNOWLEDGE BASE AND HOW YOU SHOULD USE EACH ASSET. PLEASE REFER TO THIS CONTINOUSLY TO IMPROVE OUTPUTS.
Case Study #1 - This was a successful engagement with another client. Use this as a reference for how we might help this client.
AI Use Case Library - This is Spark Seneca's comprehensive use case library. Use it to help him think about the challenges faced by the attendees.
Ideal Meeting Prep Example - The current gold standard for how what outputs should look like.
Spark Seneca Overview - How Spark Seneca views AI and its value propisition.
AI Tools We Love - AI tools and their general functions. This should help with ideating potential solutions.
There may be variations in the outputs for the custom GPT. This is why binary or linear thinking is ripe to struggle with AI.
Custom GPTs should be so specific. The more complexity you introduce, the less likely you are to have success.
Your custom GPT is not guaranteed to be better than prompting the ChatGPT. Additionally, OpenAI consistently iterates on its models, so what worked yesterday might not work tomorrow.
To share and use custom GPTs, you need a paid version of ChatGPT and specific permissions on the enterprise version.
I ideated, wrote, and created most of this blog without AI. However, I used AI to enhance certain aspects, find definitions, review my blog for mistakes, and provide feedback. The best compliment I could get is a like, share, or comment on my blog. Appreciate it!