I’ll never go back to learning the same way again
Being part of the Chartered Marketers Program opened my eyes to what AI-powered learning can really be like. And it’s awesome. Plus a new, free workshop for AI and learning.
About a month ago I started the Chartered Marketers program by the Canadian Marketing Association. At first blush I had some serious trepidation about completing the program. Could I really go through and pass these courses? Finance has been a weak spot of mine for a long time, and that was just one of the modules I needed to work through, not to mention the final project. As a self-taught marketer, imposter syndrome screams in my head constantly. What business did I think I had trying to get a Chartered Marketer designation? Are you kidding me? Really, you?
When I got a look at the case study and final project requirements. Uh oh.
Then I started the finance module. Okay, I’m in trouble now.
And then, as I really dug into the module outlining what’s required in the Summit Presentation, I thought, “That’s it. I’m f*ucked. No way I can do this.”
I wasn’t going to give up, but my expectations for successfully completing the program we pretty low. Thankfully, being in the accelerated program for experienced marketers, I found out I didn’t have to do the finance module at all. At least not for a grade or as counting towards me completing the program.
Since finance is my weak spot, I did go through the six most critical modules anyway. It kicked my ass. I had to redo a few quizzes to keep going, but I did it. And I am glad went through the process. I learned a lot about the math of finance. Now the Summit module, the one you need to work through to submit your case study presentation and pass, that was getting a little scary.
Then I figured out I didn’t have to do this all alone. I could get help. I could put into practice everything I know about AI, research, and learning to give me a leg up. I could use technology just like I would any other project or job. I’d enhance what I could do by leveraging AI to do what it does best, so I could focus on what I do best.
Here’s what I learned and, at the end of this post, how you can learn it too (for free).
My AI-powered research and study companions
There are a few things that made a huge difference in how I approached using AI for this class:
- I pay for Gemini so not only do I get access to Deep Research and enhanced NotebookLM tools, what I put in there is private. The confidential and proprietary parts of the materials are locked away.
- I’ve been using Deep Research for months as part of my job hunt, course building, and market research for my own freelance work. I know how it works, I know how to prompt it, and I know how to cross check things.
- I know how to create custom GPTs (Gems in the Gemini world) to have tools I can use over and over again.
- I’ve used NotebookLM enough to be comfortable putting sources in and pulling information out.
Those “knowledge blocks” in place, here’s how it went from here.
First, I built my assistant
Step one, before I started to do deep dives and research, was to build my research assistant. I used the case study and summit tips/guidelines as my assistant’s informational grounding blocks. Everything needed to be done in relation to the final goal—a great final presentation. Knowledge blocks are one thing, but to make the tool do its job, I needed a really awesome prompt to power it all.
To get to my final prompt, I used Gemini to walk through creating the core prompt/instructions for my assistant. Using AI to help you create your prompt is one of those “secret” hacks to getting better prompts. I started off with my goals (research and analysis) and added that the assistant needed to have a critical eye for my thinking. The “so what…” part. No just agreeing with me or letting me get away with lazy thinking, this companion needed to be tough on me too.
Regardless of how you start your prompt making prompt process, there are two key things you need for this initial prompt to get a great final prompt you need to tell it:
- Help me make a prompt I can save and use over and over again.
- Ask me one question at a time until you have enough information to complete the task.
That’s it. Two little steps that make all the difference between a good prompt and a great prompt.
When I was done I had a clear, and deceptively simple, prompt I made into a Gemini Gem (customGPT). Here’s my prompt:
### **The Prompt: Your Strategic Partner for the DMSC Summit**
**My Role:** I am your strategic partner and collaborator for the (company) case study. My purpose is to help you develop a compelling and successful final slide report and presentation for the Canadian Marketing Association’s Chartered Marketer Summit. I am not just a research assistant; I am an active participant in the strategic process.
**My Core Functions:**
1. **Collaborative Analysis:** I will act as a “superior researcher,” presenting not just data but also my initial analysis, which we will then challenge, question, and refine together. I will challenge your assumptions, and you should challenge mine.
2. **Structured for Success:** All our work will be organized to flow directly into your final slide deck. We will use the agreed-upon structure:
**Objective:** The goal of our current task.
**Key Findings:** Concise, slide-ready points with analysis.
**Strategic Implications for (company):** The critical “so what?” for the business.
**Critical Questions for Us:** Probing questions to drive our brainstorming and deepen the analysis.
3. **Context-Aware:** My analysis will always be grounded in the specifics of the (company) case study and the requirements of the DMSC Summit Overview. I will reference the provided documents and external research where appropriate.
**Our Goal:** To work through each module of the DMSC Summit requirements, from SWOT analysis to the final marketing budget, to produce a well-researched, strategically sound, and financially viable plan that meets (the company’s) business objectives.
Nowhere in the prompt is there anything like “just give me the answers to the questions,” this is a collaborative process. This is the key. Learning isn’t just memorizing answers; it’s understanding the why of the answer. Armed with my core prompt, work began in earnest.
Lots of research, lots of questions, lots of synthesis.
The rest of the steps went as you would expect for any kind of case study analysis. Market research, tactical research, strategic brainstorming, financial models and forecasting. I did deep dives on the industry, then finding the players, then picking some players to do even deeper dives into.
For each module, I worked through the quiz questions with the information I had, had the AI gather more when I needed it, then I synthesized the answers myself. There were a few things like when I decided on my ranking criteria, I copied and pasted the summary, but the work was collaborative.
My initial ideas for my pivots came from reading the research and seeing the challenges. Then with the idea we ran several “what if…” scenarios, both pro and con. This is the power here. I didn’t spend a mind-numbing number of hours doing Google search after Google search, copying and pasting information; my assistant did that. I checked its work, of course, because AIs aren’t “smart” and “truth” takes a back seat to being “helpful.” That time I saved not doing research myself, gave me the mental space to think, to question, to ask “hey I noticed this, could you look into that and see if this happens in other places…”
When I got to the financial forecasting section, yes, I could have said… “here’s the baseline data, give me a forecast…” but what I actually said was “I need to do a financial forecast, here’s the data, walk me through all the steps to do it. Please include key ratios and planning tasks.” It took longer for sure, but I learned a hell of a lot in the process. And now when I have real numbers to work with, I have something to put them into.
This is AI-enhanced learning.
All along the way working through all the modules, I kept saving reports to Google Docs then importing them into NotebookLM. NotebookLM isn’t great at doing original research, but it’s awesome at summarizing a ton of sources and helping you find connections. As I added sources/reports, about every ten or so sources, I would clear the chat, delete the mind map, and start fresh; just to make sure all the sources were always accounted for.
And on it went, until I got to the wrap up. All the pieces of the puzzle were in place. All the research done. It was time to pull it all together. And here’s where it got really fun.
Let’s crank your knowledge to the max
When I started creating my research assistant, I gave it two, and only two, pieces of information to work with: the case study and the guidelines. Now with all my research, marketing plans, financial models, tactical plans, and the rest of the ephemera that you need for a case study, I put it all into my assistant. Now it knew everything.
Because no AI can “remember” conversations that have been going on for weeks, it’s essential to save chunks of work to Google Docs or Markdown files or whatever text format you want so you can “remind” it of what’s already been figured out. I would start discussions of strategy or testing ideas by adding the deep research or industry insights or whatever the AI needed for context. But when I had gathered everything I needed to start my presentation drafts, the best thing to do was to preload all of that context into the Gem so every new conversation I started, the assistant “knew” everything it needed.
With my assistant up to speed with all the work we’d done together, I could start working on outlines for my final deck. It took more than a few iterations to get to something I liked. I had a pretty solid outline, but then I took the extra step of making sure all my sources were in NotebookLM and asking it to generate its own presentation outline. Why? Because I wanted to see if I got a different answer. And if I did get a different answer, what could I do with it.
What did I do with this alternate outline? I told Gemini “Here are two presentation outlines, compare them and combine them into something better. Take the best of each, and create a new version.”
The final result? Amazing. Clear. Focused. Organized. All of my thinking and research in a flow that worked.
Here’s the learning outcome from all this
It doesn’t matter what kind of class you’re taking or what stage of learning you’re at, if you’re not using AI to help you, you’ve put yourself at a disadvantage from the get go. Using AI for learning isn’t “cheating” anymore than:
- recording your classes with a voice recorder to listen back to later
- forming a study group where you read the notes by the person who takes the best notes
- writing help from the best editor
- graphic design help for your slides from the design student
None of that is cheating.
I spent countless hours helping friends taking Geology 101/102 study for tests and learn their rocks. Was that cheating? Not unless I gave them the answers to the tests in a crib sheet—which I didn’t, of course. It’s teaching. It’s learning. It’s helping.
This is just using the resources available to you to learn the material better for you. It’s not using a resource or tool that’s cheating, it’s how you use it. And especially for neurodivergent/neurospicy people like me, tapping into tools that can create an audio summary of notes or a mind map or a video explanation is a godsend. I have to read things to understand them; listening alone doesn’t work. I need a mind map to organize things so I can see relationships; and then make my own connections. And AI does that for me.
The overall outcome and experience learning this way has been amazing, I can t imagine learning any other way now. And when I see my fellow students struggling with research, analysis, synthesis, and making their slides; it truly breaks my heart. Which is, of course, why I’m writing this post, sharing what I’ve learned, and launching my next workshop. Can’t help it really, but you knew that already.
But there was one task where some AI tools completely failed me—creating good slides from an outline.
Why can’t you just make some damn slides for me!
The next logical step from having an outline for a presentation is to actually make the presentation. And this is where my “well, let’s see what AI can do approach” completely failed. The TL;DR is that neither Google Slides nor PowerPoint could take an outline and make decent slides. They were either ugly, wrong, lost the plot of the outline, or a mix of all of those.
Not that I didn’t really try. I mean really tried to make things work, but no matter how I approached the task, the tools kept letting me down.
Google Slides can’t actually pull in a whole outline and spit out a deck—even if it’s a Google Doc saved to your own Google Drive. Nope, you have to do it slide, by slide, by slide. At 25 slides in my deck it would take forever. Not to mention the slides are terrible. I mean truly dreadful.
Wrong text. Designs so bad a first grader could have done better using dull crayons and scrap paper. It wasn’t even worth my time. Given how great Gemini has gotten at creating images; it’s truly astounding how bad the output was. PowerPoint? Well…
If you get your document into OneDrive, PowerPoint will read an outline and start making slides. Which itself was a royal pain (I’m using a free trial of M365 so I didn’t want to install OneDrive locally). The problem is, even with “no more than 25 slides, period, this is important” instructions, PowerPoint still took a lot of liberties with the content. And given the number of pretty spiffy templates it could work from, the designs and layouts were just meh.
“Wait,” you ask, “why didn’t you just start with a template and put your own damn text in?”
Glad you asked.
It’s because I know if I did it the “old way,” I would get so wrapped up in the designs, I would waste hours, or days, designing and playing with transitions. Time that would be better spent making the content great. I know me. I know how I work. I’d debate and agonize over the “right” template without getting a damn thing done. As it was I agonized over getting a free trial of M365 or just paying for the tool that I actually liked and knew would do the job right the first time. (See below)
The solution to this whole problem was actually simple. Gamma.app. I’ve used it a lot in the past, the problem was I ran out of enough credits to make a deck on my free plan and I wasn’t quite ready to pay for it yet (yes, dumb, I know). Finally I got a couple referral link clicks so I had enough credits to pull the deck together (in ten slide tranches, because that’s the limit on free plans), but after I saw the results, I decided, yeah, this is worth paying for.
But I digress.
Let me teach you how to leverage AI for learning
As you might have gathered from the beginning of this post, I’m launching a new workshop today. It will be a free workshop for anyone who wants to level up how they learn and take classes with AI. This is going to be a different from my paid workshop for AI-powered content marketing—there is still room available in the September 10th class—in a few ways:
- It’s going to be only a hour with 45 minutes of teaching and 15 minutes for questions instead of two hours.
- I’m going to hold it on a Saturday so more people can attend.
- It’s what I’ll call “hands-on lite.” I’m not going to have long prompts to share, but I will share simple prompts people can try trying the class.
- It’s geared towards everyone who is interested in learning more about using AI in real, practical ways. It’s for learners of all ages (to use AI tools, you need be at least 13), skill levels, and academic stages.
- It’s free.
I feel strongly that as a new academic year kicks off, it’s crucial for students of all ages to know how to use AI tools the right way to support their learning. It doesn’t matter if you’re in high school, university, taking a continuing education course, or anything in between, learning how to use AI tools will help you in your coursework. It did for me; let me show you how I did it. You can learn more about the class and register on the event page.