Social Media Audit GPT: How I Built It & How To Create Your Own GPT for Work or Learning.

In integrating AI into my courses, I’ve had experience using Custom GPTs. They can be very beneficial over broad AI use as they focus specifically on a single task or project to help the user – whether student, professor, or professional. For example, I have used JobsGPT in a previous blog post as a way to help predict how AI will impact the skills marketers need in the future so that I can adjust my course material.

I was also recently inspired by an article in Chronicle of Higher Ed. In “Teaching: Can AI actually help students write authentically?” Beth McMurtrie shares how Jeanne Beatrix Law, director of composition at Kennesaw State created a custom GPT Writing Guide Assistant. She found a way to engage students with AI to teach critical thinking and the writing process through prompting versus having AI write for students.

I also realize my students need to gain experience working with AI such as custom GPTs and agents to prepare for today’s marketing jobs. The latest CMO survey reports use of generative AI in marketing increased by 116% since 2024 – now 15% of marketing activities. As a hopeful sign, the same survey reports companies are still growing their marketing teams – 5.3% last year and predicted 5.0% in 2025.

My Social Media Audit GPT. Available now – an AI assisted social media strategy tool.

 

The Primary Goal of My GPT

My goal in creating the Social Media Audit GPT was to provide students with a learning assignment to teach step-by-step an important course concept. Social media audits are an amazing strategic tool but students often struggle to understand them completely – even with new examples in the 4th edition of my Social Media Strategy book.

The custom GPT takes them through the process of completing a main assignment through prompting, and they can ask questions at any time along the way. It also has the benefit of focusing on source materials to ensure accuracy.

To create the Social Media Audit GPT I gave it an article I wrote on this blog last year detailing the process for conducting a social media audit with a social media audit template. I see the custom GPT as a great support to in-person instruction giving each student access to how I would tutor them in this key concept 24/7. For those using my Social Media Strategy text in classes, this is a great supplement to support your instruction.

Social Media Audit Template To Improve Social Media Marketing Strategy.
I trained the GPT on the Social Media Audit template from my Social Media Strategy book.

Secondary Goals of My GPT

A secondary goal was to show students how to use AI responsibly to empower their learning, not harm it. Creating a custom GPT is a key demonstration of AI integration and teaching AI literacy versus AI bans labeling all AI use as cheating. This helps teach responsible AI use for students tempted to use AI to complete assignments.

Another secondary goal is to teach students how to work with AI as a partner in developing marketing strategies. The GPT is not a replacement for those creating a social media strategy for an employer or client. The AI agent doesn’t complete the audit.

I instructed the GPT to not collect data for the user it to prompt them to formulate their own insights. The real value of a social media audit is getting into each social media platform and seeing what’s happening with your own eyes. I built the AI as a strategy development assistant demonstrating how students or professionals can use custom GPTs and AI agents in their future or current marketing careers.

How I Created The Custom GPT

I had a working model of this Social Media Audit GPT several weeks ago as a Microsoft Copilot Agent (AI-powered assistant), but it was stuck inside my institution – you can only share with individuals or groups in your organization/company. Google Gemini Gems (custom AI experts), and Anthropic Claude Projects (curated sets of knowledge) have similar limitations in that your custom AI agent, gem, or project can only be shared internally within your organization.

Only OpenAI’s custom GPTs can be published on the open web and mobile app to be shared publicly. Anyone can use Custom GPTs with a free ChatGPT account, but to create a custom GPT you need at least ChatGPT Pro (at $20 a month). Before this, all my AI use was limited to only models and tools that I could access for free (so my students wouldn’t have to pay).

Yet with custom GPTs, I was in the opposite situation. As Marc Watkins explained recently, while OpenAI and Google are giving away premium subscriptions to students, they have not extended that offer to professors. I finally secured some funding to purchase a ChatGPT Pro account.

One thing I like about my blog is I own it and control what is published there. With this GPT I’m relying on OpenAI to host for me. If I downgrade to a free account, I can’t access it. Thus, I’m locked into paying $20 a month to manage and update. OpenAI, if you’re reading, please extend the free Pro account to educators, not just students.

GPTs Are Essentially Good Prompts

What is a custom GPT? OpenAI says “a version of ChatGPT for a specific purpose.” MIT Sloan explains, “Custom GPTs are helpful AI tools tailored for specific domains or contexts. GPTs differ from standard chats through ChatGPT due to custom instructions and the ability to keep a knowledge base in addition to what ChatGPT has been trained on. This allows users to create a custom GPT to address a specific need that might be hard for ChatGPT to achieve on its own. The process … requires no code, and involves using specific prompts and your own data to provide insights into a particular field.”

AI Prompt Framework Template with 1. Task/Goal 2. AI Persona 3. AI Audience 4. AI Task 5. AI Data 6. Evaluate Results.
AI Prompt Framework Template for writing good prompts – what you need to create a GPT.

Creating a custom GPT is essentially writing a good, detailed prompt that users of the GPT will begin a chat from that background and knowledge. In creating my Social Media Audit GPT I wrote a long prompt explaining what I wanted it to do following my AI Prompt Framework of Task/Goal, Persona, Audience, Task, Data, and Results.

In the image below I marked up my GPT prompt to sections of the AI Prompt Framework. The text on the top was my original building the Copilot Agent and adjustments. The text in the bottom right is the adjustments I made in custom GPT.

Custom GPT and Copilot Agent prompts to create Social Media Audit GPT.

Test Your GPT To Make Changes

An important part of this process is to test your GPT as a typical user to see how well it performs. If you find something wrong simply tell the GPT what you like and what needs to change. You can test it in a Preview column next to where you instruct the GPT.

One of the first adjustments I made was to clarify that I wanted the GPT to have the user visit each social platform and report results. An earlier version searched the web and reported back its analysis. I tested the social audit GPT with a running brand (see below).

I like to run so I chose to test Social Media Audit GPT with Saucony running shoes and apparel

Once you’ve tested the GPT you’re ready to publish! Click the “Create” button in the top right. Then click “Share” at the top right. In that pop-up screen select “Only me,” “Anyone with the link.” or “GPT Store access.” After choosing GPT Store your GPT will be available at https://chatgpt.com/gpts for anyone with a ChatGPT account to access. Search by name or click “My GPTs.”

The custom GPT you make is only limited by your discipline knowledge, the data you provide, and the strength of your prompt.

Have you explored creating a Copilot Agent, Gemini Gem, or Open AI Customer GPT? How might you use this in your teaching for professional practice?

Please try the Social Media Audit GPT and share any feedback you have. A great feature of custom GPTs is you can revise and update.

This Was Human Created Content!

How Will AI Agents Impact Marketing Jobs & Education? See Google’s AI Reasoning Model’s “Thoughts” And My Own.

AI image generated using Google ImageFX from the prompt “Create a painting depicting the British army in red coats as AI robots coming into town to take people's jobs." https://labs.google/fx/tools/image-fx

In my last post, I warned of the AI agents coming to take our jobs like Paul Revere warning of the British coming. Large language model companies like OpenAI, Google, and SAAS companies integrating AI are promising increased autonomous action. Salesforce has even named their AI products Agentforce, which literally sounds like an army coming to take over our jobs!

Whether you’re in marketing, advertising, PR, or communications or a professor in these areas it’s important to remember AI agents and new reasoning models aren’t magical or human. They’re simply really good prediction machines. But they’re so good AI will increasingly take parts of our jobs now and potentially replace entire jobs in the not-too-distant future.

But they’re not good at everything and not always right. That’s why you need to be involved in determining how AI will be used in your job. Don’t let AI happen to you. Make AI work for you.

AI image generated using Google ImageFX from the prompt “Create a painting depicting the British army in red coats as AI robots coming into town to take people’s jobs.” https://labs.google/fx/tools/image-fx 

Productivity gains are already happening with AI.

Ethan Mollick, author of Co-Intelligence: Living and Working with AI recently shared a study that found 30% of U.S. workers are using AI every day and that it is tripling their productivity (reducing a 90-minute task to 30 minutes). If you are not in that 30% there is still time to catch up. Honestly, as much as I write about AI and implement it in my classes I don’t use it as much as I should for my everyday tasks.

That’s why I turned to Gemini for help with this post. I wanted to test a new reasoning model and see how it thinks but also use it as a research assistant. Writing an article like this takes a lot of time. In addition to testing the new Gemini “reasoning” model, I was looking for time savings in researching how AI agents may impact marcom jobs.

In this post, I look under the hood to see how AI crafts responses by seeing what Google’s new reasoning model “thinks” about the future of marketing related careers. Will AI agents take our jobs? If so, how soon? For my test, I gave Gemini 2.0 Flash Thinking a prompt that I know worries many in my field. Below is my prompt. I wanted a brutally honest assessment.

I asked Google’s reasoning model Gemini 2.0 Flash to give me a brutally honest look at the future of marketing jobs and how they will be impacted.” From https://aistudio.google.com/
I asked Google’s reasoning model Gemini 2 Flash Thinking to give me a brutally honest look at the future of marketing jobs and how they will be impacted. https://aistudio.google.com/

What does AI think about AI agents taking our jobs?

First, let’s get to know the reasoning model I used. Google explains it by saying, “the Gemini 2.0 Flash Thinking model is an experimental model that’s trained to generate the “thinking process” the model goes through as part of its response. As a result, the Flash Thinking model is capable of stronger reasoning capabilities in its responses than the Gemini 2.0 model.

How do you see its thinking? In the screen capture above you have an option to click on “Expand to view model thoughts” before you read the response. I did this to see its chain of thought and include the thought process in the screen capture below.

Gemini took a 10-step process to get to the final answer:

  1. Acknowledge the User’s Need
  2. Frame the Initial Message
  3. Structure the Timeline
  4. Brainstorm Areas of Impact (Current & Future)
  5. Assign Percentage of Impact – Now (Base Reality)
  6. Incrementally Increase Percentages Over Time
  7. Directly Address Jobs Replacement – Hard Truths
  8. Focus on Skill Sets Needed for Survival and Success
  9. Maintain a “Brutal but Constructive” Tone
  10. Refine and Sharpen Language.
I asked Google’s reasoning model Gemini 2.0 Flash to give me a brutally honest look at the future of marketing jobs and how they will be impacted.” From https://aistudio.google.com/
Google’s Gemini 2 reasoning model showed me the thinking process for responding to my prompt. https://aistudio.google.com/

Seeing AI’s thought process and its self-correction.

Before my brutally honest prompt, my prompt was to get an honest, yet reassuring answer to the question. In the screen capture below you can see how numbers 1 and 2 in the thinking process varied from above. I imagine that is how I think when writing for different audiences. That is why tools such as personas are great for marketing professionals crafting content.

In that first prompt, I also saw how it “self-corrected” in the process. An initial prediction of AI automating 50% of marketing content within a year was second guessed as Gemini talked to itself saying “That’s likely too high and broad. AI can automate some content creation tasks like basic … but not complex storytelling, brand voice development, or strategic content planning.” The self-correction resulted in dropping the number to 20-30%.

I asked Google’s reasoning model Gemini 2.0 Flash to give me a brutally honest look at the future of marketing jobs and how they will be impacted.” From https://aistudio.google.com/
Gemini 2 Thinking showed how it self-corrected a prediction about AI taking on 50% of content marketing tasks next year. https://aistudio.google.com/

Let’s get to its final response. How worried should marketers or communications professionals that support marketing be? What should we be doing to prepare ourselves and our students for this AI revolution?

The response is broken into three “Brutal Truths.” From my research and study, most of this feels accurate. Honestly, much of the first category is already happening and has been done for years by other forms of AI. So it is not surprising to me.

Brutal Truth 1: Some parts of your job will be replaced and some jobs will be eliminated.

Below is a screen capture of Gemini’s response. It predicts 5-20% of tasks will be outsourced to AI in an “efficiency overhaul.” It includes mundane repetitive tasks, basic content creation, customer segmentation plus lower-tier performance reporting and analytics. This fits what I know.

In the last two years, we’ve seen more basic content creation being done by AI whether through LLMs like ChatGPT or AI integrations into SAAS platforms such as Owly Writer in Hootsuite. For customer segmentation, I can see AI helping with data collection, but overall segmenting audiences requires more human insight.

The final one isn’t surprising. Creating auto-generated reports off previously set-up dashboards has been around for years. The important part is knowing what KPIs are important – the realm of a seasoned human strategist. A new aspect may be auto-generating initial language around the reports and a prompt overlay. But I would not rely on AI to understand the full context.

I asked Google’s reasoning model Gemini 2.0 Flash to give me a brutally honest look at the future of marketing jobs and how they will be impacted.” From https://aistudio.google.com/
Google Gemini 2 Thinking’s brutally honest truth one about the future of marketing and communications jobs with AI. https://aistudio.google.com/

Brutal Truth 2: The demand shift is dramatic. Adapt or fade.

Below is the screen capture of Gemini’s second brutal truth. The demand shift will be dramatic. It tells us to “adapt or fade.” After the brutal message, it does try to reassure us saying that marketing isn’t going away. But don’t feel too reassured because Gemini follows up with an all-caps pronouncement that it will change RADICALLY.

You want to position yourself in a high demand area. This includes strategic marketing visionaries (AI-augmented), creative directors and brand storytellers (AI-guided), and data-driven insight interpreters and storytellers. It includes AI marketing technologists and integrators, ethical AI marketing guardians, and human-connection and empathy experts. I feel confident in these areas and confident teaching students these higher level skills.

They don’t surprise me. My revelation came when I stopped thinking of AI as all-or-nothing. The scary AI agent redcoats became more manageable when I broke my job into tasks and reclaimed my human agency to decide what to use AI for and not to use it for. That’s the purpose of my AI Use Framework.

I asked Google’s reasoning model Gemini 2.0 Flash to give me a brutally honest look at the future of marketing jobs and how they will be impacted.” From https://aistudio.google.com/
Google Gemini 2 Thinking’s brutally honest truth two about the future of marketing and communications jobs with AI. https://aistudio.google.com/

Whether you follow my framework or not, I encourage you to break down your job into tasks and find the things that can be automated by AI. You’ll be surprised at what you won’t mind giving to AI to spend more time on what you enjoy more. You’ll also discover things that could be automated but should be kept for humans because the goal is to build relationships and relationships can’t be automated.

The high-demand future list looks accurate. They’re uniquely human-based skills even if parts become AI-augmented or AI-guided. The key is to make this shift yourself now. If you don’t AI will become the thing that happens to you, not the thing that you help shape and influence. Find tasks that can and should be outsourced to AI and start using it. But don’t trust it for everything. No matter how confident it sounds, it doesn’t always get everything right. Use your discipline expertise to discern and verify results.

Brutal Truth 3: Upskilling is not optional. It is survival.

The third brutal truth reinforces what I said above. Upskilling is not an option. AI innovation is coming quicker than any other technology revolution. You can’t opt out (unless you’re retiring this year). Thus, you need to become AI literate, focus on strategy and creative thinking, embrace data, learn to work with AI, and specialize strategically.

I’m not a historian or war expert, but I’ll make a final connection to the theme of my last two posts. Some factors that contributed to the colonists winning the American Revolution include being familiar with their home territory (your discipline), strong motivation (defend your livelihood), and fighting for something they believed in (human ability and agency).

The Continental Army also moved away from traditional methods of battle. Your discipline, whether marketing, advertising, PR, communications, teaching, or something else, may have a long tradition of doing things a certain way. Now’s the time to find new ways to remain relevant to keep humans in the loop during the AI revolution.

I asked Google’s reasoning model Gemini 2.0 Flash to give me a brutally honest look at the future of marketing jobs and how they will be impacted.” From https://aistudio.google.com/
Google Gemini 2.0 Thinking’s brutally honest truth three about the future of marketing jobs with AI. https://aistudio.google.com/

I’m trusting AI for the predictions, but I’ve studied AI since 2022 and they seem accurate. They also match a similar prompt I tried in Anthropic’s Claude 3.7 and what SmarterX’s custom GPT JobsGPT 2.0. predicts. I shared JobsGPT with my AI use framework to help break down jobs into tasks to outsource to AI. A new feature forecasts AI jobs by industry, profession, or college major by job title, description, and skills required – helpful for professors’ curriculum and professionals’ upskilling.

I asked JOBGPT 2.0 by SmarterX to forecast new jobs that could emerge for marketing majors as AI reshapes the industry from https://chatgpt.com/g/g-wg93fVwAj-jobsgpt-by-smarterx-ai
I asked JobsGPT 2.0 to forecast new jobs for marketing majors as AI reshapes the professional field. https://chatgpt.com/g/g-wg93fVwAj-jobsgpt-by-smarterx-ai

In the end, I feel good about what I’m doing in my classes. I’ve always focused on higher-level strategic thinking and creativity focused on human insight and emotions through storytelling. Now I’m teaching students how to integrate AI into marketing, communications, and learning tasks. What can you do to help prepare for this future?

I asked Anthropic's Claude 3.7 to forecast how marketintg related jobs will change with AI agents and make recommendations for professors. https://claude.ai/
Anthropic Claude 3.7’s forecast on how marketing-related jobs will change with AI agents and recommendations for professors. https://claude.ai/

This Was 50% Human Created Content!