Ready to unlock the full power of Gemini? Stick with me for the next 20 minutes, and I will walk you through everything you need to know—from nailing your prompts to generating killer images, creating your own custom Gems, plus a few insider tricks that will take your experience to the next level. So fire up Gemini, follow along, and let's dive right in!
Just hop on Google, type in "Gemini," and click the first link that comes up. You're probably already logged into your Google account, so we're halfway there. Now, make sure you've switched to Gemini Advanced.
Just like with other LLMs, Gemini has a couple of versions: 1. 5 Flash and 1. 5 Pro.
Flash is the freebie—solid but basic. It's got a smaller context window and simpler language processing. So if you're looking to get the full experience and all the bells and whistles, you'll want the advanced version.
This upgrade is part of Google One, which also includes extra Google Drve storage. It's basically a two-for-one. There's also a 1.
5 Ultra version in the works, though it's not available to the public just yet. When it is, it will be priced similarly to ChatGPT at $20 a month, and with the added Drve storage, it's definitely a deal worth waiting for. Now, let's break down where everything is and what to click on.
Here, the layout is pretty straightforward. On the left, you've got your recent chats, and right under that, there is a button to start a new chat. Below that, you'll see a section called Gems.
Think of these as custom GPTs. If you want to learn how to make those, check out our ChatGPT guide. Some Gems come pre-made by Google, and you can use them right off the bat.
In just a few minutes, I'll walk you through creating your own Gem. Then there is a button to manage all your Gems in one place. You'll also find help and activity buttons, which are pretty self-explanatory.
Under settings, you can tweak your extensions, share chat links, change your theme, or manage your subscription. Now, here's my first tip: head over to Extensions and make sure you've got everything selected. In this section, you'll see a list of apps and services that work with Gemini.
You've got the full Google Suite: Sheets, Docs, Drve, and so on. There's also Flights for booking hotels, Maps for directions, and YouTube for video searches. Each toggle opens up new sources of info and powers up Gemini's capabilities.
I suggest turning all of them on if you want to unlock the full set of features. Personally, I only leave YouTube Music unchecked. As for the chat window, it's super basic.
You've got a prompt bar at the bottom, plus a button to add files and one for dictation. Gemini handles multiple file types since it's multimodal, but when you click "Upload," the image option will only pull up image files on your computer. If you choose "Upload Files," it won't recognize photos, which is a bit odd.
However, there is an option to connect files from Google Drve, so you can quickly reference your own files and add data when you need them. And guys, you often ask me where are the links to all the tools I show. To make your life easier, we created the website a.
me, where we post most of the tools we show in videos. We test them all ourselves and then write a detailed review. The link is once again in the description.
Let's talk about prompting. With any language model, how you phrase things can make or break your results. There's a saying in AI: "garbage in, garbage out.
" Basically, if your prompts are vague or not specific enough, you're probably going to get a random or useless response. But don’t worry! You don't need a Ph.
D. in prompt engineering to get this right. Just follow a few simple rules.
First rule: make sure each prompt includes one of these three things—style, formatting, and/or restrictions—and always be as direct as possible. For example, if you want to know how to change the oil in your car, a prompt would be: "Explain how to change the motor oil like I’m a complete zero in mechanics. " Here, the phrase "a complete zero in mechanics" sets the style, so Gemini will adjust the explanation to be super beginner-friendly.
When researching for some videos, we in my YouTube agency often use this trick to come up with a clear and entertaining line. This makes our videos more fun to watch and more engaging. Another cool thing about Gemini is the tweaking option right below each response.
You can make the answer longer, shorter, or even change the tone a bit. No need to rephrase your prompt for that. Next tip: specify your preferred formatting.
You can say "format it as a funny essay" or whatever style fits what you're after. And don’t forget about restrictions; these are key. The more you use Gemini, the more you'll notice certain patterns in its responses.
Add restrictions to avoid unwanted styles or tones, like "don’t use academic language; keep it simple. " And one more thing: no need to follow a strict structure in your prompts. Just like ChatGPT, Gemini understands natural language pretty well.
Now, it's more about the actual info you're feeding it than nailing down specific parameters or keywords. Here’s something people often overlook when prompting: context for LLMs. Context is everything!
It's like giving the AI a backstory that helps it deliver exactly what you need without tons of back-and-forth. For instance, I could say, "I'm a first-year physics student with a 3. 5 GPA, and I have an assignment to write a short article on the Doppler effect.
Can you give me 5 to 10 bullet points? " Right there, Gemini already knows how complex or detailed to make the response. answer, you obviously don't need a link to Gemini, but if you see it in our videos and can find it, check out our website, aim.
me, where we post most of the AI tools we talk about in our videos. More context means less guesswork for the AI and keeps it from wandering off-topic. And don't forget, you can always, always ask Gemini to dive deeper into its response—just follow up with a question, and we'll keep the context from the previous answer.
Try role-playing; when you ask Gemini to act as an expert in the field, it immediately adjusts its tone, pulls from the right knowledge sources, and tones down any hallucination or made-up info. For example, if you prompt it with, “You are a biology professor, and I am a final-year student trying to understand CRISPR-Cas9 gene editing,” you are going to get a technical, in-depth response that's focused and accurate. Adding that I'm a student sets the tone.
If I change it to my 15-year-old nephew or my 5-year-old nephew, the AI still delivers the right info, just simplified to match that level. Long story short, context is king. In my YouTube agency, we never forget about context while prompting; that's how we manage to keep our results consistent because consistency is what you want—not only from AI but also from yourself, especially if you're starting out on YouTube.
When I first started, it was really difficult to stick to a schedule, especially since I was doing everything on my own. There was no AI to help me; I wanted a team back then, and I have it now. Let me tell you, the team is what builds channels.
You can try doing it on your own, but it can take years to see the results. Instead, it's better to find a good team as early as possible, and my team can become your team. In my YouTube agency, we professionally help channels grow and improve financially.
We help at every stage of producing content—from content planning and coming up with video ideas to scripting and designing thumbnails to get clicks. People in my agency have many years of experience and know exactly what to do and how to revive and grow any channel. In the video description below, you will find a link to a small questionnaire—answer a few questions, contact us, and together, we'll conquer YouTube.
There is still a lot to dig into. If you want to get top-notch responses from Gemini, here's a tip: try giving it examples of exactly what you're after. You can paste it in the sample text format, a doc, or whatever helps set the target.
Let's say I ask Gemini to write some emails using the example I provide, and boom, it whips up 10 different versions— all similar but slightly unique. Now, it didn't automatically add dates or names from my example, though. Why?
Because I didn't specify that part. Remember, Gemini, like any LLM, is a robot, not a creative director. It does exactly what you tell it, nothing more.
So think of it like giving instructions to someone who's completely new to the task; don't assume it will read between the lines. This is the main problem with AI and algorithms these days—they just don't understand. YouTube algorithms, for example, also use some machine learning and speech recognition to decide which videos to push, and in my agency, we often use other AIs like Gemini to polish our scripts, getting rid of anything that might trigger the alarms.
That's why your videos are getting decent views all the time on all our channels. Here's a trick that seems simple but is surprisingly underused: PR chaining. Imagine you've got a massive task for Gemini; maybe you want a detailed marketing plan, a full document translated, or even an entire story draft.
That's a lot, right? If you try to cram it all into one huge prompt, you will likely end up with a confusing result. Instead, break the task down into smaller steps.
Each step has its own clear prompt, and you use the output from one step as the input for the next. It's like a chain of instructions that builds toward your final goal. Sure, AI can handle big jobs, but even it can get tripped up when overloaded with info.
Chaining keeps things manageable and leads to way better results. Let's say I want Gemini to help me with a YouTube video script; first, I would ask it to analyze an existing script, including the outline and my writing style. Once it's done, I will have it create a fresh outline.
The outline Gemini suggests to me now mirrors the structure of the original script I shared, and if I need to tweak it—adding or removing sections—I can do that right there before moving on to the next step. Finally, I will have Gemini write the full script based on the outline we just finalized. If I tried to do all this in one giant prompt, trust me, the results would be messier, and yes, I've tried.
Here's a bonus tip: ask Gemini to think through the task step by step. This works across all LLMs. When it's prompted to think through a task in steps, it's much less likely to veer off track, keeps the logic tied, and helps prevent hallucinations in the response.
Simple but super effective. The same goes for reasoning; asking Gemini to consider arguments for and against something gives you a well-rounded answer that covers multiple angles. Often, it will bring up insights you wouldn't get in a straight answer.
Small tweak, big impact. Another big thing to keep in mind with Gemini is that, unlike ChatGPT, Gemini is online all the time. You don't even need to ask it to use the web; it's already doing that, so if you want to check the latest news on.
. . something like: "What are the latest updates on Chad GPT?
Gemini will automatically pull info from recent articles, even giving you links to the sources. For example, Gemini is quite handy for checking stock information. Now, I wouldn't recommend taking this investment advice as gospel, but when it comes to pure numbers, Gemini's data is really on point.
Try something like, 'Find the five best-performing stocks right now and for each, give me projected earnings if I invest $11,000 for five years. ' Just like that, you've got a list of stocks with projections. But Gemini's web skills go beyond just browsing.
Did you know it's also a powerful SEO tool? You can drop in a link and ask it to analyze the SEO keywords on the page. My go-to prompt is simple, but you can make it as detailed as you want: ask it to rank keywords, check search volume, and so on.
This makes SEO search and optimization way easier. You can even have Gemini suggest better keywords for ranking, or go all out and get it to write a full blog post. One thing to watch out for, though: Gemini's browsing isn't foolproof.
Not all websites work as expected, especially those with paywalls or cookie pop-ups. So if it gets stuck, that might be why. And by the way, we use Gemini to help us polish the blog posts on our site with AI, to aim for masterful content.
Let me show you one more of Gemini's potentials. The first thing you probably haven't thought much about is information extraction. Check this out: your task is to extract model names from machine learning paper abstracts.
Then I tell it exactly how to format everything, what to do if it hits a snag—basically explain it like I would to a total newbie. After I give it an abstract, Gemini nails it in one go. That's just to start, though; let's take it up a notch: visual analysis.
Like I mentioned, Gemini is multimodal, which means it can handle images too. I might ask, 'What do you see in this image? ' Gemini doesn't just recognize objects; it picks up all the text, identifies elements, and gives a detailed breakdown.
Text recognition is not even a challenge at this point. Here's a real test: I've got a math problem with a solution. There's typed text, handwriting, and a drawing involved.
This is a tough one, so I ask, 'Try to reason through the question step by step. Is the answer correct? If not, please explain the error and how to solve it correctly.
' And sure enough, Gemini spots the mistake and corrects it. The only downside is that it's not great at displaying formulas properly, unlike Chad GPT, so for math-heavy tasks, there's still good room for improvement. In my YouTube agency, we often like to ask AIs for help with thumbnails.
Visual analysis is a powerful thing, and sometimes the suggestions are really helpful. These improved thumbnails do help our videos get more views—we've tested that already. But Gemini's visual analysis goes further.
For example, let's say I upload an image with a bunch of subplots that need rearranging. I will prompt it with, 'Write code to rearrange the subplots in this figure using the latest version of Matplotlib. ' Then I lay out exactly how I want the plots organized and ask for a visualization of the new layout.
Gemini has to recognize each graph, understand the required arrangement, identify any necessary form, and write the code. After a bit of tweaking, it delivers the rearranged figure. Pretty impressive, right?
Seriously, I could spend hours showing off what visual analysis can do with Gemini. This is one of its coolest tricks. For example, I can upload a picture of some yarn and ask, 'What can I make with this?
' Gemini will come back with a whole list of knitting ideas, and it doesn't stop there. If I want, it will even show images of finished items so I can get a real sense of what I could make. Or think about this: here in Gemini as a cooking buddy, I just show it some ingredients and ask, 'How can I make a veggie omelet?
' It will whip up a full recipe, give me step-by-step instructions, and even let me upload progress photos to get specific guidance as I go. I'll admit it's not something you'll use every day, but knowing Gemini can handle even these more complex, multi-step processes is impressive. Plus, with the voice dictation feature, you can talk to it instead of typing, which makes it super hands-free.
Now, Gemini's skills go even deeper; it's also a data analyst in your pocket. Say you've got a CSV file or a whole database—don't waste time trying to dig out patterns or correlations yourself. Just load it into Gemini, and with the right prompt, it will do all that work.
Remember to use prompt chaining here. Data analysis can get messy fast, so break it down into simple tasks for Gemini to tackle one step at a time. Start by asking it to check out the file and understand the content.
Once you're sure it's got the info down, ask it to perform the analysis you need, like creating graphs, finding trends, or spotting any patterns, and it nails it! You can ask for any kind of graph or analysis style, and it handles it like a pro. You don't need to overthink your prompts; just keep them as clear and direct as possible.
And coding? Gemini is good at that, too! Whether it's bug fixing, optimizing code, or debugging, it's all about the same approach as data analysis: direct prompts and one task at a time.
No need to overwhelm it with everything at once; that way, it smoothly handles each issue, helping you refine your code without a fuss. For programmers, especially for the new ones, learning how to code with Gemini would be invaluable. " So easy!
I wish it was as easy with everything. I wish Gemini was smart enough to grow YouTube channels, though; that would have put my YouTube agency out of business. What really matters on YouTube is your team—your human team.
In my YouTube agency, I've assembled some of the best teams with years of experience on YouTube. These people can maximize the potential of the channel, make it popular, and help it earn more. Sounds exciting!
There's a link in the description that will take you to a small questionnaire. Fill it in, contact us, and let's crush algorithms together. [Music] Not only can Gemini analyze images, but it can generate them too, and this is where things get fun.
Gemini's been playing catch-up in the image generation game, so the prompts still come with a few quirks. But once you get the hang of it, it's pure magic! Here's a general flow: start simple, like "generate an image" or "create an image," and then just describe whatever you want to see.
Picture yourself as an artist, but instead of a brush, you're painting with words. Instead of just "sunset," go for something like "create an image of a fiery sunset over a deserted beach with pink-tinged clouds," and bam! There is a high-def, picture-perfect sunset.
The secret is giving more details. Try something like "generate an image of a cheerful shark wearing a baseball cap, eating a hot dog while watching a baseball game—one team's in orange and black. " Nailed it!
All the elements I mentioned are right there in the image. Want some action? Go for it: Gemini's all in.
Whether you're asking for a hummingbird darting around a meadow or a swirling vortex of code, Gemini has memory, so you can tweak your image anytime. Change the style, aspect ratio, resolution—whatever you need! Like, "recreate this image in the style of an impressionist painting," or "do it in Picasso's style, with watercolors or oil paint effects or specific brush types.
" It's all on the table! There are a few prompt tricks to keep in mind, though. If you're after a specific resolution, start your prompt with it: "create a 4K image of a vintage typewriter in the soft glow of a sunset.
" Don't save it for the end of the prompt. For that extra real feel, add "photo realistic" to the prompt: something like "generate a photo realistic image of a medieval fantasy meets cyberpunk fashion show, full-body view, electric blue outfit. " You will get exactly what you're picturing!
Honestly, Gemini's image generation has leveled up big time, so if you're choosing between this and ChatGPT for logos, collages, mood boards, or just about anything visual, Gemini's got you. The most useful feature of Gemini that you absolutely have to master is creating your own gems. Think of gems like custom GPT models tweaked to do one task and one task only.
It could be a writer's assistant that nails your style every single time or an idea generator that's always on point. Now, Gemini does come with a few ready-to-use models. There is the writing editor for perfecting grammar, style, and structure; the brand brainstormer that throws out ideas like it's a creativity machine; a career guide with personalized career advice and growth plans; a coding partner to help with everything from writing code to debugging; and a learning coach that breaks down complex stuff into bite-sized steps.
Just to make script writing more efficient at my YouTube agency, we have our custom gems set up. These gems polish the text, making it more natural and engaging, easier to narrate. But if you need something more specific, creating your own is a breeze!
Just follow me: hit "new gem" and give it a name. Now, unlike ChatGPT, where you might type out a whole description, here you're just filling in the blanks. But don't worry, Gemini's way is just as good.
Then start writing instructions for your gem using the prompt framework: persona plus task plus style plus algorithm plus format plus restrictions. You can mix these up in any order, but having all of them makes for a top-notch result. Here's an example: "You are a professional scriptwriter.
You write detailed, coherent, and informative scripts in a lighthearted style, and all your advice is actionable and accurate. The topic for each script will be provided first. Create an outline, wait for feedback, then continue writing.
Format it as plain text, without comparisons or imaginative phrases. " This is already a solid prompt, but if you want to boost it, click on the magic wand. Gemini will improve your prompt, making it more detailed and specific for the LLM to follow.
It basically builds on what you've written, so the more detailed your prompt is, the better the upgrade. That is just a matter of testing, tweaking if necessary, and hitting "save. " And that's it—your custom gem is ready for action!
Everything I've shown you so far is just a foundation for your new skills. You are past the basics now, and with a bit of practice, you will only get better, especially since learning Gemini has never been easier. Just remember: be as descriptive and direct as possible.
Treat it like a robot that knows everything but needs you to spell out what to do. Push it—don't hold back! Try new prompts, set up fresh tasks, and don't limit yourself to what I've shown you here because, with Gemini, you can do so much more.
Give it a bit of time, keep experimenting, and soon enough, you will be an AI master yourself. Thanks for watching, and I'll catch you in the next video!
