27 FREE Things You Can Do With Gemini

Transcript

00:00

CocaCola glazed onions. I'm going to try that.

This was a very worthwhile deep research in my opinion. While ChachiPT and Claude seem to be adding more and more expensive tiers for you to be able to use, Google seems to be making more and more cool features just available

00:16

for free. Now, this video is in no way sponsored by Google.

I've just found myself using Google's AI more and more often because they just keep on baking in cool stuff into their platform. So, check this out.

Gemini made me this game

00:31

with just a one prompt. This game is called Emoji Fusion Master.

I didn't name it. Gemini did.

But it's basically the same concept as that infinite craft game where I can take two items, for instance, Earth and Moon. Click fuse

00:48

item and then what it's going to do is it's going to discover a new item, give it a new name and then actually use AI to generate an image to represent that item. So for example, when I fuse those two items together, it gave us a Terra Luna orb.

And that little image that it

01:05

generated, well, that was generated with AI as well. It used the Google imagining model to generate this image.

I built this game completely for free inside of Gemini. You can actually see I just minimized it.

It's inside of Gemini. Here's all the code that it wrote.

Here's all the files it created. And I

01:22

gave it the prompt, build a unique and fun game that has never been made before, but is super fun and addicting. It should also be colorful and beautifully designed.

I didn't even give it an idea for a game. I said just come up with an idea for a game.

Now, I can't really say this hasn't been done before. Infinite Craft already exists, but it

01:39

made this game in one shot. Now, if you want to do something similar, maybe you have a little bit better of an idea of the actual game you want to develop, again, head over to a studio.google.com.

When you get there, you'll see a page that looks like this. Over on the left

01:54

sidebar, there's an icon that says build that looks like a little puzzle piece. If we click this icon here, you can enter an idea for any game or any app you wanted to make.

And most of the time, it will do it with just one prompt. Just for fun, let's have it create another game.

This time I'll give

02:10

it a little bit more details. Create a game similar to Frogger, but instead of a frog, you're an alien and instead of cars, you're dodging spaceships.

Make it colorful and beautifully designed. I'll go ahead and submit this.

And we can see that it's just going to go off and start figuring out what it needs to do. Right

02:27

now, it's putting the whole concept together. Once it's figured out what it's going to do conceptually, now it's actually starting to write the code.

And after roughly 2 minutes here, it created Alien Dodger. I can click on this little full screen button here and view the game in full screen.

This is my first

02:43

time testing this. So, I'm going to go ahead and enter the level.

And you can see my little alien down on the bottom. And I guess those are spaceships.

And it works first. Oh, I got hit by a spaceship, but it worked on the first try.

Now,

02:58

obviously, you're probably going to create more complex games than this. I just wanted to do a quick demo and if I was to make a more complex game, it would have taken me a lot more time.

I'm definitely going to do that in a future video though. But it's not just good at creating games.

It can also create useful productivity apps for you as

03:16

well. For example, I use a tool called Feedley to stay on top of all the latest AI news.

I plug in a bunch of RSS feeds from a whole bunch of tech blogs and resources like Microsoft and Google and things like that to just really stay looped in. Let's say I want to create my

03:32

own version of this and get away from Feedley. I don't actually want to get away from Feedley, but let's just say for example purposes I did.

I could build a custom app just like this for myself this time. Let's do it through a screenshot though.

I'm going to go ahead and screenshot what Feedley looks like

03:48

here. Now I'm going to jump back over to AI Studio and I'm going to add files.

So I'm going to go ahead and upload an image here and I'll grab this screenshot of Feedley that I just took. I'll then add the additional prompt.

create an RSS feed reader that looks similar to the screenshot. I'd like to be able to add multiple RSS feed sources and have the

04:04

app pull in the latest news from that feed. I'd also like to be able to sort feeds into folders.

So, very similar to what the very basic model of Feedley does. Let's submit that.

Once again, it is going to analyze what it actually needs to create and really think through it. Once it's done analyzing, it will

04:21

then start writing all the code. Once it's done writing all the code, it actually does a quick bug check on itself, fixes any bugs that it finds, and then gives me the final version.

We can see that it built our app over here on the left side. Again, I did this all with one prompt.

You can see my original

04:37

prompt here. It thought for 40 seconds about what it should create.

Then it worked on the code for about 218 seconds and it analyzed a bunch of errors, fixed its own errors, and now we've got, let's go ahead and pull this up into full screen. something that looks just like

04:53

Feedley. Let's go ahead and add an RSS feed real quick.

So, I can click this button. We get a little popup.

I'm going to go ahead and grab the RSS feed from ZDNet. Probably one of my favorite tech news resources.

Jump back into my Feedley clone here. Add our first feed.

05:08

Let me just paste that in here. Add feed.

Look at that. It pulled in all of the news from ZDNet.

Just like that. Pretty dang good for a first try out of the box.

I made meat a fee clone and it looks pretty much identical. Now, also in the AI studio, another feature that I

05:26

feel like doesn't get nearly enough love, but people could probably get a ton of value out of it is if we come over to stream, we've got options for talk, webcam, and share screen. You can actually share your screen and an AI will walk you through how to do

05:41

something specific that you're trying to accomplish. Let's say I'm trying to edit a video in Da Vinci Resolve, but I'm still learning Da Vinci Resolve.

So, let's go ahead and click share screen. I'm going to select entire screen.

I've got two monitors set up, so I'm going to select screen one here. We'll click

05:56

share, and then it's going to start the chat. Can you help me make some updates in Da Vinci Resolve?

Yes, I can definitely help you with updates in Da Vinci Resolve. What specific changes would you like to make?

I want to remove the background here and just keep the person in the video. How do I get rid of

06:12

the background? Okay.

To remove the background, you can use the magic mask feature in the color page. First, please go to the color page.

Then, select the clip in the timeline. After that, look for the magic mask tool, usually in the power window or qualifier panel, and use

06:29

it to draw a stroke on the person. Okay, I think I've done that.

What now? Now, you need to track the mask.

Look for the tracking options, usually around the magic mask tool, and click the track forward button to follow the person's movement throughout the clip. You might

06:44

also need to track backward if the clip starts later in the action. Okay, I just did the tracking.

Now what? Now you need to isolate the person.

You might need to invert the mask so that the background is selected. Then you can apply effects or adjustments to the background to remove it or make it transparent.

This

07:02

might involve using a layer mixer node in the color page. Can you just look at my screen and tell me what to do?

Okay, I see the magic mask applied. Now, please look for the invert option for the mask.

It's usually near the magic mask settings. Click it to select the

07:17

background instead of the person. After that, you might need to add a new node or adjust the existing node to make the background disappear or replace it.

You can use a layer mixer to combine the masked person with either a new background or just a transparent background. Well, I guess uh that's good

07:33

enough. It removed the background, but it also didn't keep my headphones and hid my mic and everything.

But I just love this idea of using AI as this like interactive tutorial where I'm trying to learn a new tool and it can actually almost look over my shoulder and tell me where to click and what to do. You can

07:49

actually do that with AI Studio. I know kung fu.

But it's not only your screen that you could share things with. You can also share your camera and it can tell you what it's seeing on the camera.

If I click on webcam inside of the streaming tab on AI Studio, it now has

08:05

access to see my webcam so I can hold things up to the camera and ask about them. What is this thing that I'm holding up to the camera right now?

Based on the display showing thermal images and the label, it looks like a thermal camera. Possibly the Fleer

08:20

TG165-X model. It's used to detect heat signatures.

Can you tell me how hot my face is right now from the video? According to the display on the thermal camera, your face appears to be around 93.6° F.

Maybe I'm sick. The temperature

08:36

shown on the camera is 93.4° F. However, surface temperature measured by a thermal camera can vary quite a bit from your internal body temperature.

If you're feeling unwell, it's always best to check your temperature with a regular thermometer and consult a medical

08:52

professional. Can you tell me what's on my hat and also what does it mean?

Your hat says feature tools which suggests something related to advanced or innovative technology or equipment. It could be a brand name or a slogan emphasizing cutting edge tools.

Another

09:07

thing that a lot of people don't realize you can do over inside of AI Studio is you can actually give it YouTube videos and it will watch the YouTube videos. Now most people probably assume, okay, it goes and looks at the transcription of the video and then tells you about the video based on the transcription.

But that's not how Gemini works. Gemini

09:24

actually can watch the video. To prove this point, I'm going to jump over to YouTube here.

And one of my favorite channels is this Fire Ship channel. Absolutely love every video he puts out.

One of the things he does is he uses a lot of memes in his videos. So the audio, he's explaining what is going on

09:41

with whatever the story of the day is, but in his visuals, it's very, very memeheavy. So, if I was to copy the link to this video here, jump back to AI Studio and paste this video link in, you can see that it found the video.

And to

09:56

prove that it's not actually reading the transcripts, I want it to tell me information about what it visually sees in the video. So, let's say, give me a list of the memes that were shown on the screen during this video.

Obviously, information you can't get from just

10:12

getting a transcript of a video. And let's click run and see what it gives us.

Now, because it is watching a video, it uses quite a bit of tokens. So, it's going to take a little bit of time to do this analysis for us.

After about 30 seconds here, we can see that at 51

10:27

seconds. Awkward look monkey puppet.

109 and 117, how it started versus how it's going. 233 Spongebob 250 Elon Musk scam alman tweet.

Big brain wojack. I don't want to play with you anymore.

We can see we got a list of the memes. So, let's look at 51 seconds.

Awkward look

10:44

monkey puppet. Let's jump to about 51 seconds.

Here we have the weird monkey puppet on the screen. 233 Spongebob time card.

We can see it's got the few weeks later, which is a common thing you see in a Spongebob episode. 308 Big Brain

11:02

Wjack. When we jump to that timestamp, look at what we got.

So, you can see it used the visuals. It actually watched the video.

The transcript had nothing to do with the analysis that I just got from it. However, saying all that, it does struggle with long videos because of the amount of tokens.

For example,

11:18

here's an interview that Colin and Samir did with Corridor Crew, and the video itself is an hour and 22 minutes. If I was to copy the URL of this video, jump over to AI Studio and paste it in, you'll notice that it will use 1.46 million tokens.

Well, our token limit

11:36

with this model is only a million tokens. So, this video is too long for it to actually analyze what's going on in the video.

You can see my run button is grayed out because the token count for the prompt has been exceeded. However, that doesn't mean that with an interview like this, we still can't pull

11:52

out all of the relevant details if we wanted to. In this case, I actually would use the transcript because if it's watching the video, it wants to use 1.5 million tokens.

If I come over to this video on YouTube and I grab the entire transcript here from this video, we can

12:07

see it is a long transcript, but I'm going to go ahead and copy the entire thing. Now, if I jump over to AI Studio, let's get rid of this video and let's paste the transcript in.

You can see now it's only using 22,000 tokens cuz it uses a lot less tokens to analyze text than it takes to analyze video. So, now

12:24

that I gave it this entire transcript, I can add an additional prompt to the top that says something like, "Brown down all the key takeaways from this video into bullet points." We'll go ahead and run it. And 39.6 seconds later, I have a bulleted breakdown of pretty much all of

12:41

the main takeaways from this video. I still love listening to podcasts like this.

I actually did listen to this entire episode here, but it's a good way to quickly decide, do I want to go and listen to this whole thing? Is it going to talk about stuff that actually interests me?

It's a great way to just

12:56

get that quick look at the entire podcast all at a glance, basically. Another thing you can do inside of Gemini that I don't think most people realize you can do with Gemini is you can actually have it transcribe a video for you.

Maybe there's a video that you

13:12

need a transcription for that's not on YouTube right now. It's just on your computer.

Well, you can upload that and have it transcribe it. For example, I can drag and drop in one of these videos that I recently created about the links robot.

It's going to upload it into Google server real quick. And I could

13:28

say, give me the transcript of this video along with the timestamps. It will even give you timestamps which could save you a lot of time if you do stuff on YouTube.

Let's go ahead and click run. And as we can see, 6.6 6 seconds later, we have the entire transcript

13:43

with the timestamps. Now, this was only a one minute video.

I uploaded a shorter video for the sake of speed. One thing to keep in mind is it only allows you to upload videos that are 2 GB or less.

So, if you have a really long video, it's not going to work very well, especially

13:58

if it goes over 2 GB or if it's more than, let's say, you know, 20 or 30 minutes, you're going to start to reach your token limits and it's going to be a struggle. But if you've got a video that's, you know, 15, 20 minutes long and it's underneath that 2 gigabyte file size, you can have it transcribe it and

14:15

give you timestamps all in one shot. Another thing you can do inside of AI Studio is you can actually do text to speech, very similar to what you get from a tool like 11 Labs, but you can do it right now for free inside of the Google AI studio.

If I come up to the

14:31

left side over here and click on generate media, we have a few options. We have Imagine their image generator.

We have a Gemini speech generation and a few other tools down here. But for this example, let's do a Gemini speech generation.

Not only can you do text to

14:46

speech, but you can actually have multiple speakers all in one session. I'm going to go ahead and pause and type in some stuff for it to read real quick and we'll see what it's capable of.

So, I entered some text. We could pick voices for our speaker one and two.

Let's do core. Ready to make something

15:02

amazing? For speaker number one and then speaker number two, we can use AoE.

What kind of problem could we solve? I changed my mind.

I want to use Saddle Tiger. What kind of project are you thinking of?

I could click run and let's hear

15:19

what our little mini discussion that we created sounds like. Hey, who is the best person to follow online about the latest AI tools and news?

Definitely Matt Wolf. If you're not subscribed to his YouTube channel, what are you even doing with your life?

Uh, yeah, Matt

15:36

Wolf. I love his website, Future Tools.

I'll definitely check out his YouTube channel for sure. His Future Tools newsletter is a gamecher as well.

Radical dude. Skippidity Toilet Ohio Rizzler.

Perfect. You could also do

15:51

single speaker audio, which very similar to 11 Labs. You just give it a bunch of text and it will read it out loud.

And you can give it instructions like read aloud in a yelling angry tone. We get something like this.

Who the heck is this dang Matt Wolf guy? This doesn't

16:08

have the features where you can train your own voice into it, which is something that you get out of 11 Labs. But if that's not a concern of yours, you just want to create fairly realistic audio generation, this works pretty well.

And of course, you can also generate images with it. In fact, OpenAI, their image generation model,

16:25

the 40 image generation model that everybody's doing the studio Giblly style images with, Gemini actually kind of had that sort of feature first. Chat GPT just really overshadowed it.

But you can talk to Gemini in the same way and have it generate images for you. If I

16:41

come up to my model settings up here and I come to Gemini 2.0, you can see there's Gemini 2.0 flash preview image generation. If I select that one and I give it a prompt like generate an image of a fish wearing pants.

2.9 seconds later, we have an image of a fish

16:57

wearing pants. I like how it did the two legs of the pants, but then the fin was only coming through one of them.

But similar to OpenAI, you can also edit images with a text prompt. Make the pants purple.

And just like that, we get pretty much the same image, but it changed the pants to purple. Put sunglasses on the fish.

And there you

17:14

go. I mean, it doesn't quite line up with the eyes, but we got sunglasses on the fish.

But similar to chat GPT, you can also upload your own images and edit those images. So, let me just toss in this image here of myself.

And let's do the old-fashioned convert this image to

17:30

the style of Studio Gibli. Click run.

And well, that didn't work out too well. I guess it doesn't do everything as good as JGPT, but I can do things like put a Padres's hat on the man.

And yeah, I got a man with a Padres's hat. All right, so their image editing still has a little

17:45

to beat desired. Maybe it doesn't quite beat chat GPT there, but it is free to use and you could play around with it until you get what you want out of it, I guess.

All right, for the next couple examples, I'm actually jumping over to gemini.google.com. I've left the AI studio.

Now we're on

18:01

their actual Gemini chat app, which is kind of their version of chat GPT. And just to show you that everything I'm doing is for free, I'm actually logged in on an account that is not upgraded.

You can see I've got the upgrade button here. If I click on this little dropdown, it's asking me to upgrade.

This is the basic free version of

18:18

Gemini. No special features in this one.

Now, of course, you can do pretty much all of the same conversational stuff you can do with Chat GPT. You can have brainstorm sessions with it, ask for ideas for titles or thumbnails, have it help you write emails.

All of the same

18:34

stuff you would normally use ChatGpt for, you can do inside of Gemini for free. So, I'm not going to demo a lot of that.

You've probably seen a ton of examples of those kinds of things. Instead, I want to show off some of the areas where I feel like this really excels over some of the other models.

It's actually really good at generating

18:51

graphics, things like maps and charts and graphs and and that sort of stuff. So, if I give it a prompt like create a world map and color code it based on population, so we can quickly at a glance see the most populated and least populated countries.

And then I want to turn on canvas cuz this creates docs and

19:08

apps. That's how we're going to get our map back.

And let's go ahead and submit this and see what we get. It's writing some code for us to build this interactive map.

And about a minute later, I've got this interactive map. I can actually move my mouse around and hover over it.

And it shows the population of each country that I hover

19:25

over. And it's colorcoded.

So the darker countries are the countries with the higher population. All with a single prompt.

And again, all on the Gemini free plan. Now, let's have it create a chart for us.

I'll give it the prompt, create a line graph that demonstrates the growth in social discussion around

19:41

the topic of artificial intelligence from 2000 to 2025. Again, we'll turn on canvas.

And let's run this prompt. And again, about a minute later, I've got a graph that shows the exponential growth of people talking about AI.

And if I

19:57

hover over it, you can actually see the year and it has like a number to represent the volume of discussion around artificial intelligence. Now, you can probably see how beneficial this will be if you're doing things like video explainers or blog posts that need to represent data in a visual form.

You

20:14

can easily use Gemini to come and create this kind of graphic for you. Now, another thing that I really love about Gemini is they have probably, in my opinion, the best research feature out there.

The Perplexity one's pretty good. ChatGpts is pretty good, but I don't think any of them actually come that

20:31

close to Gemini's deep research. When you do a deep research on Gemini, it will give you like a multiple page Google document of all of the research it found.

And it is just so much more in-depth than all the rest of these platforms give you. While to really get

20:46

the most benefit out of the deep research, you do need to be on one of their pro plans. They do give you a certain amount of free uses per month.

I believe you could use deep research about five times per month on the free plan. In order to do it, come up to your Gemini model at the top here, select 2.5 flash, and you'll notice that we now

21:03

have the option to turn on deep research. So, if we turn on deep research, and we tell it to do something like break down the possible toppings for a hot dog based on all of the available hot dog reviews, and we submit this, it gives us a quick game plan of

21:19

how it's going to go about that research. So, it's going to identify and gather a diverse set of hot dog reviews.

From the gathered review, systematically extract all mentioned hot dog toppings. Categorize the extracted toppings into logical groups.

And you can see all of the steps. It's going to analyze the

21:34

results and create a report for us. And it will take a few minutes to do.

Like it it takes time to do the research, but let's go ahead and approve this plan by clicking start research and let it do its thing. And now it took several minutes to complete this, but I've got my ultimate hot dog topping compendium

21:51

here, which one thing that I really love about it too is I can quickly export it to docs and it opens up the entire document in a Google doc that I can save for later. And this is one quite honestly I'm probably going to look at again.

Like look, an Alaska reindeer dog, CocaCola glazed onions, ketchup,

22:08

mustard, and relish, bacon wrapped hot dog with grilled onions, pinto beans, tomatoes, mushrooms, cheese, mayonnaise, guacamole, and green salsa. I'm going to try that.

This was a very worthwhile, deep research in my opinion. But it is interesting to look back and look at how indepth of research it gave me.

It even

22:25

broke it down into like a chart here with the regional style, the signature toppings, the dominant flavor, and we got quite a good list of hot dogs like smoky sweet pulled pork, pickled red cabbage, crushed pork rind. I'm in.

And then when we scroll all the way down, it

22:41

will give us a giant list of the resources that it found this information from. Again, I really love using this deep research feature.

And on the free plan, you get a handful a month that you can use. All right, the next thing I want to show you is not in the Gemini app.

It's not in the AI studio, but it

22:56

leverages Gemini, and that is Notebook LM. If you have never tried Notebook LM, you are in for a treat.

Basically, what it does is it allows you to upload a ton of information about a specific idea or product or research or whatever kind of information or documents you want to

23:12

give it. You can then either chat with that information, have it create a mindm from that information, summarize that information, or probably most people's favorite feature, turn that information into a podcast.

I can literally upload PDFs, markdown files, MP3 files, PNG

23:29

files. I can link it to Google Drives.

I can upload websites or YouTube videos. Or I could just copy and paste text in.

So, let's give it a website link. I've been hearing about this birds aren't real movement and I want to learn more about it.

So, let's go ahead and plug in

23:45

their about page. Now, you can enter a lot more information if you want, but just for fun, I want to see what kind of podcast it gives us around this bird's art real conversation.

So, I'll go ahead and click generate underneath the deep dive conversation option. And it will

24:01

take a couple minutes to generate, but you don't have to wait because here's what it sounds like. What if everything you thought you knew about the natural world was actually one big lie?

That's quite the opener. It is.

Yeah. Today, we're taking a deep dive into an organization that claims exactly that,

24:18

specifically about birds. They say the birds you see, they aren't real.

Yeah. We've got this source material that lays out this uh incredibly detailed history, decades of it.

Filled with government secrets, big claims, shocking revelations. Our mission today is just

24:34

to unpack it all chapter by chapter, understand the narrative they're presenting, right? Follow the story as it's told.

So, that is another great feature of Gemini that you could use for free. You can learn about any sort of document, website, anything you want in podcast form just by plugging it into

24:50

Notebook LM. Super super cool.

And then finally, probably the most talked about feature from Gemini right now is their V3 video model. It is super powerful.

It generates really, really good-looking videos, but even generates audio as well

25:06

with those videos. And to be honest, V3 is quite expensive to generate videos with.

Now, you have to sign up for their Ultra plan, which is a $250 a month plan to generate a lot of videos with VO. They do have a less expensive, I believe $20 a month plan, which gives you like a

25:23

lot smaller amount of generations, but it's still fairly expensive to use. So, for those of you that hung around all the way to the end of this video and watch this whole thing, I've got a little bonus tip for you.

You can actually generate videos with V3 for free right now thanks to Perplexity.

25:39

Here's how you do it. If you head on over to x.com, you can actually use the Ask Perplexity bot.

Let's go ahead and create a post on X. And I can tag the Perplexitybot by tagging at askplexity and then tell it what you want the video to be.

So, I can give it a prompt like, "Make me a video of a monkey on roller

25:56

skates in front of a wolf howling at the moon." And when I post this, the Perplexity bot will then turn around and make that video for me. I did exactly that.

And here's the video that it gave me, complete with audio and everything.

26:16

So, there you have it. A ton of ways to leverage Google Gemini completely for free.

You can probably see why after watching this video why I've been kind of obsessed with Gemini lately. There are so many things that you can do with it.

Now, some of the things I showed you can do with ChatGpt. And ChatGpt might

26:33

even be better at some of them. But literally every single thing that I showed you in this video, you can do right now for free with Google's tools.

A lot of this same stuff if you tried to go do it with ChatGpt, you would need to be on one of their more premium tier plans. It's really, really cool to see

26:48

that Google caught up. A couple years ago, we were talking about how, wow, Google really fell behind in AI, and Microsoft and Open AAI really passed them.

Well, now I feel like Google caught up and is pretty much level playing field with them, and they're giving some of their best tools away for

27:04

free for you to use right now. I don't know if that will be forever, but as of the recording of this video, you could do everything I just showed without paying Google a dime.

Pretty pretty sweet. But that's what I got for you today.

Hopefully, you enjoyed this video. Hopefully you learned something new and have all sorts of new ways to go

27:20

and play with Google's Gemini. Again, this wasn't sponsored.

It's just one of the tools I've been personally using most myself recently. Thanks for tuning in.

Hopefully, I'll see you in the next one. Bye-bye.

Thank you so much for nerding out with me today. If you like videos like this, make sure to give it a

27:35

thumbs up and subscribe to this channel. I'll make sure more videos like this show up in your YouTube feed.

And if you haven't already, check out futuretools.io io where I share all the coolest AI tools and all the latest AI news. And there's an awesome free newsletter.

Thanks again. Really appreciate you.

Summary

Transcript