Transcription
We don't lack AI tools. We need better system thinking. More AI tools do not mean more productivity. Instead, we need to develop deeper expertise and build better AI workflows.
So in this video, I'll share my favorite simple AI workflows using the most popular AI right now: ChatGPT, Gemini, Claude, NotebookLM, Perplexity, which will immediately improve your work productivity and quality. Let's go.
Stop asking which AI is the best because in reality, it doesn't exist. Instead, think of them as your team of specialists with different strengths. ChatGPT, the generalist assistant; Gemini, the integrated builder; Claude, the visual strategist; Perplexity, the specialized researcher; NotebookLM, the knowledge synthesizer. Now, the right question becomes, what is the right tool for this task?
For this video, I've organized workflows into three core areas: research and analysis, strategic visualization, asset creation. The goal is not about using all these tools. These are just starting points. And once you see how they connect, you will discover your own combinations.
First, we search and analysis. This is where we need to find new information and make sense of it. For most people, we often start with ChatGPT deep research because that's the tool we are most familiar with. But instead of just simply starting a deep research, there is a much better workflow. And this is my personal favorite.
Let's say we're doing market research and strategy brainstorming about a furniture brand. The first step is not to do research yet. Instead, I always recommend creating a ChatGPT project so you can start building persistent knowledge by uploading background information like a brand overview document in this case. So the trick is to run the deep research inside this project, and it would be much better as now ChatGPT has all the important context and will generate a better and relevant research report about your target market and make recommendations.
Then we can upload this report back to this brand project as part of the knowledge base. So essentially, you are training this project with all the important research findings. Now you can even prompt it to get deeper insights and generate different strategy documents. So, for example, a detailed B2B buyer profile document that contains your ideal customer profiles based on your research, or a competitive positioning document, brand voice guide. And they will be all relevant to your brand because of the project context it has been trained on. Now you can repeat the same process, upload them back to this project, and build a powerful brain that is self-improving with more useful context.
*******
And this is a strong foundation before you even start using ChatGPT for various tasks like content creation, image generation. I have another whole video on how I use this one single workflow to build a marketing system. I'll put the link below.
Now, what if you want total control and don't want AI to pick the sources for you? Here is what I love combining Perplexity and NotebookLM. Perplexity lets you curate the exact latest sources you want. And then NotebookLM only analyzes what you import with minimal hallucination. So you control both the input and the output quality.
So I'm doing competitive analysis about expense management tools. So first on Perplexity, ask it to find the top 20 expense management platforms. Note, ask it to just output the URLs only. So we can have a list of URLs that is ready to import back to NotebookLM. Then on NotebookLM, we can paste back this list and then import all of them. Now you can also configure the chat and then assign a persona to this notebook project. So perhaps a growth consultant specializing in market positioning. So all the responses will have this persona embedded.
Now let's ask it to analyze all these competitors' homepages and identify the five most repeated value propositions so we can avoid them or enhance them further to find a differentiation. So here NotebookLM will act as the growth consultant, not just providing the insights on the overused messaging, but also the strategic opportunities as I have specified in the persona instruction.
Now another useful feature is the report generation. So here you can select and generate different types of report formats. Besides the default building report template, it also will, based on your sources, suggest other formats for you, like competitor analysis format in this case. And I'll suggest keeping the instructions unchanged. It has already incorporated the relevant context for you. And then, just a minute, we have this report ready. It is super comprehensive with the side-by-side comparison. And the best thing is that all this information is now based on the latest competitor homepage sources that we have imported. And so we can make sure this is highly accurate with minimal hallucination compared to a deep research report where you can't control the source quality. And you can even just copy it and then create a Google document that is ready to send out.
And if you want to go beyond the basics with NotebookLM and Gemini and turn it into a complete system at work, I recommend this guide from HubSpot, Google Gemini at Work. This guide shows you how to leverage the Gemini marketing stack like deep research, NotebookLM, Gemini model, image generation for marketing tasks, research, strategy, content creation. It even includes a four-week implementation plan that walks you through from quick wins to full team adoption and system integration, so you are not just trying to figure out everything at once. My favorite part is the prompt templates included in each section, like for strategy, there is a complete strategy template that is ready to use, or the content multiplier template for content creation. You can just copy them and customize for your own needs. This will be way easier than starting from scratch. Download this in the description below for free, and thank you HubSpot for sponsoring this video!
Now, part two, strategic visualization. This is about turning research information into something we can see and use for decision-making. And here is where Gemini becomes a powerful tool. Its massive context window can load entire research reports into the canvas and build ready-to-share visualizations in seconds. And so, combining with Perplexity search, you now have a powerful research-to-visualization workflow.
Let's say I'm doing audience research on Perplexity about a service business. So first, ask it to find discussions about the biggest frustrations, purchase considerations from real users when working with marketing agencies. So Perplexity has different specialized search modes, and this is what makes it useful. So we can pick the social search mode to find the latest Reddit discussions, and we can even turn on the deep research mode for deeper insights, and it would give you a lot of insights about the user challenges. And then we can just export it into a PDF report.
Now on Gemini, upload back the Perplexity report and then turn on the canvas mode and ask it to open this research report. And the beauty is, because Gemini has a large context window, so it will load the exact report into the canvas. So this is amazing. And the reason for this is, Canvas will then give you access to the building creation tools on Gemini for infographics and other visualization creation. And from my experience, this canvas method will always generate better infographics than just prompting in a chat. Now you can see there are lots of beautiful charts and diagrams, bite-sized visualizations that tell the story, and even process diagrams, which is great. And the whole infographic is ready to share with a link. A bonus tip is, if you really like this visualization, you can just prompt Gemini to add a "download as image" button. And so now you can just one-click to save it as an image. And this works perfectly, and you don't need to scroll and take the screenshot yourself. And if you want, you can even add an "export to PDF" button as well.
Gemini is great for quick visualization, but what if you need something more strategic? And that is where Claude excels. From my experience, Claude always generates more thoughtful, precision-driven visualizations. And here is the best thing: Claude projects let you save dashboard templates as reusable assets. So, combining with Perplexity, you can have a powerful, reusable workflow for visualization creation.
For example, I'm doing trend analysis about the tech market. So again, first, we can go to Perplexity and then ask it to find some trends reports. Note, I have used the search operator "filetype" to just filter those PDF report results, and then look through the sources and pick those you find credible, and then download it. Now go to Claude and create a new Claude project. Let's name it as "Market Analysis Dashboard." Now upload back the report you just downloaded and perhaps any color palette that you like. And then ask it to create an integrated dashboard to visualize these report insights and with the areas we want to cover. And just within a minute, the dashboard is ready. And this is what I like about Claude: it can always generate strategic charts and dashboards. Like in this case, and what I really like is, it is also using the color palette that I have specified.
Now the best thing is, you can just click here and then add this whole dashboard back to this project. And that means now we have this dashboard template as part of the Claude project knowledge. And then we can even set up the custom instructions for this project to let Claude know that this project is designed for generating dashboards. Then the magic happens: whenever you have another industry report that you want to build a dashboard, so you can just upload it back to this Claude project, and it will generate the same formatted dashboard to you every time.
*******
All right. Here is the part three, asset creation, and that is how to turn the core ideas into different asset formats. And here's another unique Claude feature most people miss: SVG artifacts. SVG means fully editable vectors, so you can change the colors, fonts, everything in a design tool like Figma. And no other AI tool does this so well, like Claude.
So perhaps you want to create a social media post about the latest Canva World Tour updates. Then you can just ask Perplexity to quickly summarize the 10 key takeaways from the event. And that is the reason why I always stick with Perplexity, because it is really good at getting real-time search and it is quick as well. Then just copy the response. Now on Claude, paste back the response in the chat, and then ask it to visualize these takeaways and turn it into an SVG artifact. Note, this is important to specify. It is an SVG artifact in the prompt because otherwise Claude might just create a typical HTML dashboard.
And again, Claude will always create beautiful infographics, like in this case. And unlike Gemini, these SVG files are downloadable and editable. So to edit these SVG visuals, you can just use any UI design tool like Figma and customize whatever you want. So the background, the font size, the title, the elements. So once you're done, you can just export it into an image directly on Figma. And so pretty quickly, you now have this ready-to-use social media graphic. And so you don't need other AI social content tools. You can just turn any insights into social visuals by just using Claude. And personally, I use this a lot so that you can always focus on your thinking and let Claude be your visual designer.
Now, let's also talk about presentations. And this is one of the top use cases when it comes to asset creation using AI. And the great news is Gemini just made a big update. You can now create full slide decks using the canvas mode and directly export it to Google Slides. And this is huge. But there is an even better workflow that is combining with Gemini GEM. So we can semi-automate the whole process.
So first, let's create a new Google Gemini GEM, which will be used there to generate a presentation outline. So let's name it as "Presentation Outline Generator." Very important is under the instruction input. So include what this GEM is used for and the workflow details when generating the outline. So now we have these new GEM created. So let's try using the deep research report from the previous example about the furniture brand, and immediately Gemini will generate the presentation structure together with a slide-by-slide breakdown for this report. And then we can just turn on the canvas mode and then ask it to create the actual presentation using this outline. And with around two minutes, the presentation deck is ready. And I haven't specified any styling or color scheme, but for my testing, Gemini will always pick the theme and style based on the materials context that you provide, and it would source the images that would fit into this presentation. So it looks pretty decent, and it's definitely presentation-ready with different images, icons, and layouts. Although it's not like those fancy presentations from other AI presentation tools, but it is very promising, and I believe it would just keep improving.
Now the best part is, it has the built-in download function where you can just one-click to export it into a PDF file. And even more. You can also export it into Google Slides, and so you can make more changes. And I must say, this is one of the best recent updates from Google Gemini.
Now we can take this to the next level. That is to build a full asset package. Sometimes just presentation slides are not enough. You need videos, toolkits, decks, multiple formats for one topic. And that is why we need NotebookLM. It now supports video overview generation, other than podcast-style audio. And combining with Gemini, you can build complete asset packages in hours.
For example, you're preparing an internal team training about the topic "Data Storytelling." So first, you can use the "Discover Sources" feature within NotebookLM to find authority sources. So I use NotebookLM here over Perplexity. That is because it prioritizes authoritative sources, so making it better for research requiring high accuracy. And then we can source data, storytelling frameworks and principles, best practices, and video lectures from authority sources. A very important part is to review these sources before you import to ensure the source quality. So especially a subject that you are not familiar with.
Then we can first ask it to design a course curriculum about data storytelling, and then it will give you a detailed course structure design and the details for each module. And this is a great start, so make sure you save it as a note. Rename it so you can always reference back. Then ask it to generate the module one details and do the same. Then convert this module one note as a source. Now to generate the video lessons. Select only this module one source and click "Video Overview." So here you can pick the format, language, what topic to focus on, and even the style. But I personally still prefer the classic theme. And then after a few minutes, the video lessons are ready, and it looks pretty good using the module one details as a source. So using this method, you can continue to build training materials. And if you really want to try a different style, you can just pick the one you like, and you can even set a custom instruction to give it a persona, like a data storytelling trainer in this case. You can see the result is quite impressive, and we immediately turned the same module into a completely different style.
But we are not done yet. Now we can even ask NotebookLM to prepare the lesson toolkit materials, including the hands-on exercises and self-assessment checklists. Then copy this response. And on Gemini, we are gonna build the materials. So turn on the canvas mode and then ask it to generate the presentation for this lesson toolkit. And now you can see every slide is following the toolkit instructions from NotebookLM. And this is a great complementary asset to be used along with your training video lessons. And you can even ask Gemini to build other learning tools to make your training package even more complete.
NotebookLM and Gemini is a powerful combo, but it is not limited for learning assets only. You can use the same workflow for business assets creation as well, like building a landing page, product prototype, pitch deck, business tool. And the great thing is, now the assets will always be backed by grounded research insights.
So let's use back the expense management platform example. So first, on NotebookLM, we can find some more sources like recent user reviews, industry reports and white papers about the market, and even some Reddit threads where users discuss switching tools. If you want, you can also use Perplexity for this step. And if you have any information about your product or service, I also highly recommend you input to NotebookLM.
Now, we can ask it to create a strategic brief for a new expense management platform landing page based on these sources. And then it will create a detailed strategic brief for this landing page, including the platform identity, core value proposition, messaging pillars. And make sure you always save this as a note. Then we can just copy the response, paste back on Gemini, and then ask it to build a fully functional, ready-to-deploy landing page based on this brief. And within a few minutes, the landing page is ready. It looks super cool and tech-savvy as I instructed. And I found Gemini is really powerful in doing this kind of rapid prototype landing page building. And the best thing is, this landing page is based on the latest market research that you have done on NotebookLM. So make sure you spend some time on the research source quality before you even start building it on Gemini. And then you can just copy this source file and then upload it back to your server hosting to publish it live.
All right. These use cases are just some ideas, and there are still lots of other ways to use them. If you feel inspired, I actually have other videos deep-diving into all other use cases and how I use them together. I'll put all the links below. And so don't let them limit your imagination. Understand their strengths and their role in your workflow design. Be creative and always think beyond that.
If you enjoy this video, please give me a thumbs up, and also I invite you to join my community. You'll get access to all the prompt resources I use on my channel videos. We also have regular live workshops, and our members are incredible and helpful. Before you go, also watch this video about Google Gemini, it's getting really powerful. I'll see you next time.