Master NEW Gemini in 33 Minutes (2026 Tutorial)

2025 have been Google's year in AI. They started slow, but then they hit their stride. And now Gemini is going punch for punch at the top tier with updates coming fast and getting better each time. Cuz reality is when Google commits, it never stays in one app. It gets integrated across their products and you truly feel it in your day-to-day tools. And for that reason, I've put together an end-to-end tutorial for you on how to use the Gemini application. And because I respect your time, I will

highlight the topics that we'll cover in the video, so you can skip ahead if you're here for something specific. If we haven't met before, I'm Ali Salam and I currently work as a director in a tech company. And on this channel, I'll help you turn tech and finance into your personal advantage.

Here we have Gemini. And this likely looks very familiar to you if you've used Chat GPT or any other AI. You have the chat in the middle, which is where you interact with the AI. You have historical chats in the sidebar to the left, as well as your gems, which we'll cover later in the video. What makes Gemini really special in comparison to a lot of the other AIS is its native integration into the Google Workspace. And honestly, I cannot overstate this. If you are natively working in Gmail,

Google spreadsheets or presentations and you're not using Gemini already, this will be a gamecher for you. And look, there are many ways that you can access the workspace function. But the easiest way to do it is literally just to do an at and then it will kind of call off any of the tools that are connected. So if you for example want to set up a meeting, you can select Google calendar and you can say schedule a meeting at 1600 tomorrow for deep work. And if you fire this away, it will call the

calendar tool and actually set up the meeting for you. Here we go. And if you want to recall anything within your Google calendar, you literally just do the same. So you can say at Google calendar when do I have my deep work session tomorrow. And there we go. I won't go through all the things you can do in the Google Workspace because that is a beast of its own and definitely out of scope for this video. But the key takeaway is that if you want to use Gemini to its fullest, then you

need to make sure to leverage the Google Workspace integration because even though most of the other AI tools can do something like this, it is not even remotely as close to being this smooth. Now, if we move on to the models, you can click the little button down here and you'll see two models. You have a fast model and the reasoning model Gemini 3, which is the latest model at the time of this recording. And I would recommend you to always keep the reasoning model on because unlike many

of the other reasoning models, it has a variable called thinking level which kind of determines the effort associated with the prompt and automatically tunes the model's effort based on it. So lower complexity tasks will go quicker whereas complex tasks will take a bit more time. Lastly, I just want to talk about Gemini's ability to consume data. It's multimodal, meaning that you can feed it pictures, documents, Excel sheets, videos, or even code. You either do that

by drag and dropping the files that you want to use into the chat, or you can type directly into the chat. You can, of course, also add the files by clicking the plus button down here, and then you would pick the source where your files currently sits that you want to kind of import into Gemini. Furthermore, Google's context window is crazy big. It can consume up to 1 million tokens within your prompt. And that means that you could essentially put multiple books

into the chat and ask Gemini to action it, which is pretty wild. And from here, we're going to start taking a look at the tools section down here. There is some really cool stuff in there. So, let's get going.

So you may or may not have heard about deep research before but let me tell you deep research in Gemini hits differently. So what most deep research capability out there does today is that it kind of goes to the web searches for a set of credible resources and then compiles a report for you which is really good. But let's now take a look at what Gemini does which is an enhanced version of that. So if we start off by opening the deep research on the tools here, you can instantly see that in your

sources, you can indeed pick the web search as a source, but you can also use the Gmail, the drive, and the chat as sources, allowing you to use a much more precise set of data inputs. But that's not it. You can also drop files in here, and that allows you to use this super powerful deep research capability in a very directed manner. So I have a very famous report here from McKenzie looks like this covering the state of AI in 2025. If you haven't read this, you

definitely should because it is really cool. And what we can do is to use that report as our input data point for the deep research. Now let's upload that file into Gemini and let's do a gap analysis to see what they potentially might have missed in the report. Now the way we want to do that is obviously we've dropped this in. You could of course have just uh clicked files, navigated over to your file directory where you have this setting and and clicked that, but I dragged and dropped

it. And then we want to toggle on some benchmark data. So what I will do here is I will let search be toggled on. So what Gemini will now do is kind of look at the report and then compare the report against other data sources to see if it can identify some elements that might have been missed in the report. Now, I'm going to drop my prompt in that I have pre-prepared. I won't bug you down with uh covering the contents of the prompt, but what I'll do is I will

it. And then we want to toggle on some benchmark data. So what I will do here is I will let search be toggled on. So what Gemini will now do is kind of look at the report and then compare the report against other data sources to see if it can identify some elements that might have been missed in the report. Now, I'm going to drop my prompt in that I have pre-prepared. I won't bug you down with uh covering the contents of the prompt, but what I'll do is I will

leave it in the description and in the pinned comment at the top for convenience. All right, let's fire this away. And by the way, if you were to do this exercise for some internal task in your company, then you should of course toggle on the sources in here where you have your file sitting. So if your company works in drives for example then that would be a good source to toggle on here when you're doing this type of exercise. Now we can see here that Google deep research as it always does

outlines the exact steps that it's going to take before executing the research. Now you can edit the plan if you want to do that. If you look at the steps outlined here and you go like nope that doesn't line up with what I was hoping for it to do then click this button over here. But to me this looks like it's directionally correct. So I'm just going to allow it to start the research. Okay. Okay, so we can see that the gap analysis report has finished and it's

built it in a very common structure. So it's starting off with an executive summary outlining the key takeaways from the report and then it goes into the specific chapters where it breaks down each topic at hand. Now typically what I would have expected in a report like this is for Gemini to have put out a table outlining the key gaps that it has identified, but it doesn't seem to have done that here. Let's scroll down and see what else it has external benchmarks

in it. So here we can see that it has started to take a swipe at the Mckenzie report. So the energy and sustainability crisis the gap the PDF contains zero references to energy consumption, electricity shortages, nuclear power or the environmental impact of AI scaling and then it breaks that down further. And then we have shadow AI and the dissolution of the perimeter etc. So you get the point. It kind of breaks it down and honestly seems to highlight a couple of very interesting

in it. So here we can see that it has started to take a swipe at the Mckenzie report. So the energy and sustainability crisis the gap the PDF contains zero references to energy consumption, electricity shortages, nuclear power or the environmental impact of AI scaling and then it breaks that down further. And then we have shadow AI and the dissolution of the perimeter etc. So you get the point. It kind of breaks it down and honestly seems to highlight a couple of very interesting

points in here. And by the way, if you haven't read that report and you are into AI, I would highly recommend you to do it. It is one of the biggest reports on AI that came out this year. Okay. The last thing I just wanted to show you on the deep research capability is now that you actually have the report, you have notebookm capabilities attached at the top here, which basically allows you to create a web page on it. You can build up some infographics. You can run

quizzes and flashcards if you're trying to learn the contents of the report. Or you can have an audio overview which allows you to listen to this report as a podcast which is really cool because it allows you to digest a report that contains a lot of complexity in your preferred medium of consumption.

Think of gems as your personal mini experts that you can save and trigger whenever you want to. And they're actually really simple to build. All you do is click the gem section up here and it will bring you into the gem manager and you can see my old gems down here and you can create new gems by clicking the button here. You have four inputs. You have a name of the gem. You have the description of the gem and then comes the important part. This is the instruction. So here is where you tell

the gem what to do. Lastly, you have a field called knowledge here at the bottom where you can add files for reference when the gem is firing off. And once you have done this, you can click save and you can also try it out as a preview in the right hand side field here. But instead of building up a new one, let's just look at one of my previous gems. So we can use this executive gem. I'll just click edit here and you can see how it looks from the inside. So obviously the name is then

executive gem. The description is that it turns text into executive style briefs and the instruction covers what the gem should actually do. So you can see here that I've given a role executive communication specialist and the task at hand is to take any input text and transform it into a concise, polished and executive style brief. And then you have a set of guidelines down here. And really this is all it does. Let's use a cute example here. Tell me the story of Little Red Riding Hood. So,

executive gem. The description is that it turns text into executive style briefs and the instruction covers what the gem should actually do. So you can see here that I've given a role executive communication specialist and the task at hand is to take any input text and transform it into a concise, polished and executive style brief. And then you have a set of guidelines down here. And really this is all it does. Let's use a cute example here. Tell me the story of Little Red Riding Hood. So,

we're asking it to give us the story of uh Children Tail, but it's going to use the instructions in the gem to produce the executive outline. So, as you can see here, it started off with subject, operational postmortem logistics security, breach, and asset recovery. All right, this is going to be the best version of Little Red Riding Hood you've ever seen. So, we have the executive summary again, uh, context, problem, analysis, resolution. So, you get the point. It can basically take any input

and turn it into an executive brief. Now, this essentially means that you have a little mini expert that you can trigger anytime you want to create an executive brief on whatever topic that you're working on. And anytime you trigger the gem, you will end up having a separate chat that you can continue working on. So you can see here little red riding hood security something. So if you want to continue working on this executive brief, you have the ability to do so. Quick pause. I have a favor to

ask. If you're enjoying the video so far, you should consider becoming a part of a small but very exclusive group of around 5% of viewers that have subscribed. And if you've already subscribed, I just want to say thank you. You're the reason why this channel keeps growing and keeps getting better.

ask. If you're enjoying the video so far, you should consider becoming a part of a small but very exclusive group of around 5% of viewers that have subscribed. And if you've already subscribed, I just want to say thank you. You're the reason why this channel keeps growing and keeps getting better.

Now, when it comes to video and image generation, Google provides some of the best tools out there. And a lot of people even say that Google are leading in this space. So, let's start off with video generation. In Gemini, the video generation is powered by VO3.1, which generates high-quality video with sound. Now, let's try that out by creating a marketing video for a hypothetical cafe. And you would do that by clicking tools, and then you select create video. And you can see here that

it's highlighting VO3.1, which is the latest model at the time of this recording. And then you describe your video. So, I will dump my prompt in here. And again, I'll spare you the time of going through it, but I'll share all of this in the description and pin it in the comment. And then all you do is fire it away. All right, here we go.

Okay. So, it's not bad. It actually followed the prompt fairly well. It gave us three sequences, right? So I gave it a zero to 2.5 second sequence. Micro shot of coffee steam rising over a cup. Followed that fairly well. And then we had a second shot. And then we had a third shot. Now I did notice that it did not say stock of Sweden, but instead it said, let's take a look here. Stock. Now I don't know where that is, but I would like to go there. When it comes to

images, let's pop open a new chat. Gemini can both create images that are stunning, but then it can also edit pictures using Nano Banana Pro. And it's doing that extremely well. So, let's say that we wanted to build a menu for our cafe and stock. Then you would need to again go into tools and then click create images. You can see that they've even tagged this functionality with a banana for nano banana. And then we'll drop in our prompt again. I'll share this and let's

fire it away. Here we go. This looks pretty good, right? Let me show you the ability of Nana Banana. If we just want to edit this picture, let's say switch the theme to forest green. All right, here we go again. This is why Nano Banana is amazing. Context continuity for video and image generation has been an issue across the board when it comes to AI generated content. But as you can see here, it literally just switched the color theme of the menu, but the rest stayed the

same. We still have the same format of the menu. We still have the same menu elements. It's still Nordic winter latte, fireside, flat white, etc. Which, by the way, looks like a really well marketed menu, right? Nordic Winter Lot. This sounds amazing. But the point is that it basically kept all the elements of the menu. We can shift back into the old one here. It just switched the color. The rest stayed the same. What this means for you is that you can keep iterating and keep enhancing an image

same. We still have the same format of the menu. We still have the same menu elements. It's still Nordic winter latte, fireside, flat white, etc. Which, by the way, looks like a really well marketed menu, right? Nordic Winter Lot. This sounds amazing. But the point is that it basically kept all the elements of the menu. We can shift back into the old one here. It just switched the color. The rest stayed the same. What this means for you is that you can keep iterating and keep enhancing an image

until you're happy with it without having to worry too much. I won't say that your worry should be completely gone because it still messes it up sometimes, but generally speaking, it's doing this extremely well and that is why Nana Banana Pro is getting so much hype because it's just really good. So the next capability is Canva which is essentially a built-in workspace in Gemini where you can take a prompt and turn it into a real output like a document, slide style content, study

guide, or even a simple web page. And the reason why it's so useful is that it keeps everything structured and editable. So you can essentially iterate on a draft similar to what you would do in Google Docs or Word or any of the other corporate tools that you're likely using today, but you don't have to regenerate a new message or a new chat every time you want to make an edit. So I have another report again from McKenzie. I'm not affiliated with Mckenzie. It's just ended up being that

these are the two reports that I picked for this episode. And this report is on the state of grocery retail in 2025. So they did this together with Eurocommerce and it was a fairly interesting report to read through. Nevertheless, let's drop that in and let's toggle on canvas under tools and we can simply say generate me an HTML dashboard outlining the key takeaways of the content in this report. open the simulation as a preview directly in the app. Here we go. So, it has built out a dashboard

for us. Starts off with an executive summary followed by a set of key data points and then it has kind of structured the dashboard into various elements that summarizes the content of the report and it looks pretty good. Now, let's say that we wanted to adjust this dashboard a little bit and perhaps take this section here with nine key trends for 2025 plus and dig a little bit deeper into it. So, what we can do is just take that and say build out a separate tab for for the section.

for us. Starts off with an executive summary followed by a set of key data points and then it has kind of structured the dashboard into various elements that summarizes the content of the report and it looks pretty good. Now, let's say that we wanted to adjust this dashboard a little bit and perhaps take this section here with nine key trends for 2025 plus and dig a little bit deeper into it. So, what we can do is just take that and say build out a separate tab for for the section.

Each element should contain a dropdown tab with further detail. And this is the cool part with Canva that you can just keep iterating like this, keep enhancing. And there is actually a button on the right hand side here which is grayed out because it's um it's currently rebuilding the dashboard. But this button allows Gemini to kind of tap into the dashboard and suggest edit enhancements. itself. Here we can see the updated version. So now there is a separate tab for deep dive. The rest

looks fairly coherent to what it looked before and it seems like it has added a button down here that takes you to the second tab. But let's shift gear and take a look at how it constructed up this uh this new section. So again, we have nine key trends. It has the drop down with some further detail inside of it. Why does this matter to you? Well, if you're building out a report and you want to skip the whole shuffle of going back and forth in the chat and just keep

working iteratively on an output like this, then Canva is going to be a great choice for you to execute that type of work.

All right, guided learning is obviously really great if you're studying or trying to learn something on the side, but it's also really useful if you, like me, work in corporate, which I know a lot of you watching are doing. So, let's say that you get sent a corporate document, a long PDF containing a lot of information that you need to learn by heart for a presentation or a Q&A, or let's just say that you want to absorb or digest that content with more ease

and without having to spend as much time doing it. What you can then do is take that report. Let's say that we got sent this state of AI report that we looked at earlier in the video. You drop that in here and then you turn on guided learning under tools and then you can just say I need to quickly understand the key concepts of this report. Where do I start? All right. So what you can see that Gemini is doing here is it says let's explore the report together.

The central theme here is a massive gap between adoption and scaling. was giving you a short introduction paragraph here and then this is key to help you get a handle of the key concepts. Where should you where would you like to start and you can now pick one of the three main areas of the report that you can dig further into. Now the reason why I think this is pretty interesting is because a common but very strong presentation technique that is especially useful when

discussing complex topics is something called scaffolded learning. And the core principle is fairly simple and you can actually see it being applied here. The idea is that you start with a very simple approach. Allow the audience to feel like they're following you as you gradually increase complexity. And you can leverage the same concept in Gemini. If you say like I did here, where do I start? Or you can also say like explain this like I'm five, the famous movie quote, you would effectively be

triggering the scaffolded learning. And at the same time, if you want a specific detail in the report, you can of course go straight into that as well. It's not by any mean, shape, or form fine-tuned to always use scaffolded learning, but rather it will adapt accordingly if you want something specific out of a report. So, let's say that we were to pick one of these. Let's say the rise of the AI agents. Let's go with 2. Then you will see that Gemini keeps unfolding further

information and dive deeper into more and more complexity as you decide where you want to delve deeper and it will always prompt you with a next step. So you can see here to help you get a better grip on this which aspect would you like to explore next and it gives you the three options. So you kind of just need to tag along here and Gemini will basically do the hard work for you. And by the way, you don't have to use a report like I did. That's just a pretty

good corporate use case for those of you that are interested in that. You could basically ask whatever in the chat, turn on guided learning, and Gemini will start training you on that topic.

The visual layout feature transforms standard text into rich magazine style pages with images, cards, and interactive widgets as opposed to kind of giving you the standard wall of text that we always get in the chat. Common use cases would be exploring complex topics or more casual things like travel planning shopping comparison or recipes. Let's say that you wanted to pick a more complex topic like learning how large language models work. All you would need to do is go into tools again,

toggle on visual layouts, and then you could prompt explain how large language models LLM work to a nontechnical audience using a visual guide. Now we fire that away. All right, here we go. Large language models, a word prediction machine. So you can see kind of these uh interactive uh pictures that are toggling through. So it says here that imagine an LLM as a highly sophisticated autocomplete function. Dude, you're not selling yourself very well, are you? It doesn't truly

understand the way humans does, but it analyzed patterns in billions of pages of text. For those of you that don't know, the way LLMs basically work is that it's a statistical model that guesses the next word in the sequence. And then it basically further down here says that there are three core steps of an LLM. So step one is training, step two is the tokenization, and step three is the prediction. And because I have fat fingers, I decided to click on step one. And it started compiling the next

steps, which is fine. I plan to show you that later. But let's just jump back here for a second. The important part here is that you get a structured output telling you how LLMs work. Now, this could of course have come in a chat as well, but this is a little bit more easy and fun to digest than a wall of text. And what it always does is give you a set of suggestions on what to explore further. So here it suggested bias and ethics, retrieval, augmented generation,

which you probably heard me harp about a couple of times if you are not a first-time listener, etc. And if you were to click any of these, it would then build out the next response. Now what I did is I clicked step one up here. So the training and now it has compiled the next element of this visual layout which is around training. And you can see here that it outlines three phases of LLM training, which is data cleaning, then pre-training, and lastly fine-tuning. So the key thing to take

away here is that if you prefer this type of output when you are chatting with Gemini, then you can just turn on visual layout and you'll get it in this format instead.

scheduled actions works very similar to the way chatbt does it in case you have tried it over there. So let's just jump straight into it. In the chat all you need to do is basically tell Gemini to set it up for you. So let's for example say create a weekly scheduled action every Monday at CT. send me a summary of the latest AI updates from past week. And that's it. You just fire it away and Gemini will set that up for you. And there we go. So basically what's going

to happen now is that every Monday at 8:00 my time, it will create a summary for me based on the latest news in the AI space and give me a notification that that has happened. So if you for example have the app downloaded on your phone then it's going to give you a notification and you will also see it in the left hand side here. You can actually see one of my um one of my other scheduled tasks which fired off earlier today and you can of course always come back into this chat and turn

it off. You can edit it here as well. But it will of course be quite painful to try and find a very old chat if you for example have had a scheduled action laying around for a while. So there's a better way of doing that which is to go down to the settings and then there is a tab here called scheduled actions. So you just go in there and you will see all of your actions listed. And from here you can either toggle them on and off like this. You can go in and adjust

it off. You can edit it here as well. But it will of course be quite painful to try and find a very old chat if you for example have had a scheduled action laying around for a while. So there's a better way of doing that which is to go down to the settings and then there is a tab here called scheduled actions. So you just go in there and you will see all of your actions listed. And from here you can either toggle them on and off like this. You can go in and adjust

the details of it. And of course, if you just want to take it out completely, you have the delete button down here.

All right, this wouldn't be a tutorial if we didn't talk prompting best practices. So, let's just quickly cover that off. For Gemini 3 Pro, which I recommend you to use instead of the fast model, it's fairly simple. It's a reasoning model meaning that you should treat it more as a kind of senior colleague rather than a junior colleague which your standard chat models like chat GP5 would be considered as. That means that you need to focus less on how it executes a task and focus more on

what it should achieve. Now Google are outlining three points that are important to be aware of when you're using Gemini 3 Pro. So the first is to be concise in your input prompts. So it says Gemini 3 responds best to direct clear instructions. It may overanalyze verbose or overly complex prompt engineering techniques used for odal models. And this is very important to be aware of because with normal chat models, the general rule of thumb is the more context the better. But that

actually confuses a reasoning model. Keep it short and to the point. The second thing is output verbosity. The key takeaway here is that Gemini 3 is by default less verbose and prefers providing direct efficient answers. And that basically means that if you want it to be more elaborative around the topic, you need to tell it to do so. So it basically says here, if your use case requires a more conversational or chatty persona, you must explicitly steer the model in the prompt. Eg explain this as

a friendly talkative assistant. Now, if you're like me and prefer the model to get to the point, then it's just good to know that this is the way it works. But you don't actually need to do anything about it. In fact, you can actually prompt it to be even more direct if you prefer that. I tend to do that, by the way. And the last thing to be aware of here is around the context management. So basically, the key takeaway here is that when you're working with a large

data set, you should place your specific instruction or question at the end of the prompt after the data context. And that is typically very different to how we have looked at prompt engineering for all the previous models where the task typically comes first or second if you have a persona but it's basically at the top of your prompt. And here it's essentially telling you to put it at the bottom and instead focus on the context management. But all in all the key

takeaway here is you don't need to overcontextualize when working with Gemini 3. It's really less is more. Now, the last thing I want to leave you with is some important settings to be aware of. So, the settings you'll find down here. And the first one, and probably one of the most important ones, is your baseline instructions for Gemini. So, if you click that, you can basically tell it to apply an instruction every time you talk to Gemini. Now, I have one, but you can

have multiple ones. And the way you add them is by clicking the add button at the top. And you can delete everything that you have on the button here. And then you can toggle them on and off using the slider at the top here. So mine looks like this. It's basically my preferred way of uh interacting with Gemini. Be concise, direct, and mei. We just looked at the um prompt engineering principles for Gemini 3. It is already very concise, but I like my responses to

be even more concise. and then basically outlining a set of rules and guidelines that I wanted to abide by as we are talking and it goes through this instruction every time we talk. So I would highly recommend you to put your version in and I'll share mine in case you want to copy the one I'm using. The second important setting are the apps and the apps are essentially the way for Gemini to pull information in context. So, in the beginning of this video, I showed you how I pulled information and

sent information through Google Calendar. That is because it's connected via this this button here. So, you can go into this space and just add your workspace in if you want to leverage the Google Workspace when you use Gemini. And then you have a set of other integrations available for you. So, you have GitHub at the top here. You have YouTube Music and Synth ID at the bottom. But as you can imagine, the most important one is going to be your Google Workspace. All right, moving on to the

next one. Schedule tasks we covered just uh just now. Then you have something called your public links. If we were to go into one of our chats, let's say that we use this older chat where we talked about how LLMs work and we want to share this, then we would click at share, this little logo down here, and it will create a link for us. So let's say that we now have this public chat that anyone with this link is able to access. If you go into your public links, that chat now

next one. Schedule tasks we covered just uh just now. Then you have something called your public links. If we were to go into one of our chats, let's say that we use this older chat where we talked about how LLMs work and we want to share this, then we would click at share, this little logo down here, and it will create a link for us. So let's say that we now have this public chat that anyone with this link is able to access. If you go into your public links, that chat now

appears. So you will have a list of all your chats that is publicly available through the link. And this allows you to both remove chats that you no longer use as well as find the the link in case you want to reshare it. And again, you have the ability to delete all links at the top here. Now, the last thing I want to show you is the theme. As you've noticed, my Gemini is dark. That is actually because I have recently switched computer and my new computer is

appears. So you will have a list of all your chats that is publicly available through the link. And this allows you to both remove chats that you no longer use as well as find the the link in case you want to reshare it. And again, you have the ability to delete all links at the top here. Now, the last thing I want to show you is the theme. As you've noticed, my Gemini is dark. That is actually because I have recently switched computer and my new computer is

set to a dark theme by default. So, even though this one is set to system, it just picks up that my system settings uh is geared towards a dark theme. Hence, Gemini is dark too. But usually when you fire up Gemini, it will look like this. So it's just a preference. I personally think that dark theme is easier on the eyes. So I like to go with this. But I also know that some of you prefer to have a light theme because it's easier to read the text. Just go for the one

that floats your boat. And now you've learned how to use Gemini. I hope this video was of value to you. Let me know if it was and let me know if it wasn't because this channel is for you. So, your feedback really matters. Now, if you want to continue your learning path in the Google ecosystem, I will point you to this video over here, which is another tutorial this time on Notebook LM, which is another Google product. And if you haven't heard about it before,

it's kind of fine-tuned to draw insights from your documents with the lowest possible hallucination. And by the way, it uses Gemini as the underlying AI model to do that. As always, thank you for trusting me with your time. And have a good one.