Weekly AI Tool Review
Every single week I try to test out just about every single
AI tool that gets released, and this week I wanted to share with you seven that
I recently found that I think you're going to find very valuable.
1. Google Illuminates
Now the first one on the list is from Google. It’s called
Google Illuminates. Now, this is very similar to another function that you'll
find inside of another app called Notebook LM. This is from Google; this is one
of their most popular and one of the most popular tools in the world of AI. So
if you haven't used Notebook LM, I'll just briefly show you what that is.
Features of Notebook LM
At notebook lm.goole.com, you could add all kinds of
sources—your own information, your own research, your own knowledge base—and
then you could have an AI-based chat conversation with all your own sources.
One of the most interesting things inside of Notebook LM is this feature called
audio overview, where it takes all that information in those documents and
creates a podcast. This was an 8-minute podcast that he created for me, and I
made a different video all about Notebook LM and this audio overview, so I'll
link that below because this is not a Notebook LM video.
Overview of Google Illuminate
So, Google decided to roll that out in a whole different
app. I'll show you that now, and by the way, if you go to labs.gooogle, this is
where Google is testing out all kinds of new AI apps. So I checked this
website; Google LM was first released here. They have all kinds of different
tools here that they're testing out, but this one is called Illuminate. Let me
go ahead and try this to show you.
Google Illuminate transforms your content into engaging
AI-generated audio discussions between two different people, using the same
voices that are inside of Notebook LM.
Limitations and User Interaction
Now down here, they also have a library. I'll just press
play on one just so you see what it sounds like: "Ready to break down some
research? This paper, Artificial Intelligence Index Report 2024, explores the
rapid advancements and growing influence of artificial intelligence (AI) in our
world."
It's absolutely incredible how fast AI is evolving to create
those audio overviews. On this page, you have to search for a topic or paste a
PDF URL. A huge limitation right now is that you can't upload your own PDFs, so
I'm hoping they roll that out. Right now, again, it's a very beta
version—actually, an experimental version of this. It says those PDFs have to
come from this link right here.
Discussion Generation
So if I go to this website, they have it on all kinds of
different topics, or you could type in a topic and press enter here, and it's
going to show you all kinds of topics that you could pull in these PDFs from.
Let me just look up this one and this one right here. I'll select these two,
I'll add those as my sources, and then over here it says "Created
discussion tailored for you."
You could change your audience, so General audience is what
you picked by default, but you could go and change that, type in your own
audience with a longer duration. It says you could do medium, you could do
longer, and you could do quick, which is less than 5 minutes. The tone should
be friendly. So in this case, let me change this to medium, and then you could
go ahead and press generate from here.
Now, this took about a minute here to create a 6-minute
medium-length audio discussion. Let me go ahead and press play: "Welcome
to our discussion today. We’ll be delving into two fascinating papers that
explore the philosophical underpinnings of physics and the nature of physical
laws."
You could press this hand signal right here, and then it
opens up this Q&A where you could ask any questions about those papers. So
a fantastic tool, except I really hope they add the option to upload your own
documents, like you could do inside of Notebook LM. But pretty cool tool
already if you find the right research papers to create these audio
discussions.
2. Napkin AI
The next one is called Napkin AI, and the website is napkin.
This will help you turn any type of text prompt into very interesting visuals
that you could use inside of PowerPoints, inside of AI-powered presentations,
and it's really easy to use. They have a completely free plan available as
well.
How Napkin AI Works
With Napkin AI, you just have to paste any type of content
that you want, then turn into a diagram or visual elements, or you could just
create by using a text prompt here. The AI actually does two things; it writes
you an entire blog post if you wanted to use this blog post, but I want to
create this visual element right here.
I’m going to go ahead and click this; it's going to create
visual elements, and look how nice this looks. This kind of graphic will take
quite a while to create, and these are other alternatives to it too. So you
just hover over each one, right? If you like any of these, go ahead and click
it, and it will save it like this. If I want to download it, let's say I don't
want to use this post at all. I'm going to go ahead and download it as a PNG.
If I want to turn any other parts of this into a visual, I
could just select that portion of it. Here's some other options that we have.
These are really nice. Let me see what other styles we get out of this one. Oh,
very cool! Select that again, download this too.
3. Gamma
This brings me to the third tool here, which is called
Gamma. Gamma I’ve covered multiple times; it's the best AI tool right now for
creating presentations, an alternative to things like PowerPoint. You just
could paste text, you could generate using a single line of text, or you could
import any type of file. I'll show you this one for now.
Presentation Generation
I'm going to create a presentation for hiring the right
people. This is the number of cards; if you have the paid version, you could
get more cards. With the free version, you're going to be limited on how many
cards—cards are kind of like slides. Here's the outline; I'm going to press
continue, and I'm going to choose a theme. Let me just choose this theme right
here, and this is what it's going to look like.
I'm going to press generate just to show you the power of
Gamma. It’s kind of one of the most mind-blowing AI tools. Look at this; it's
creating the entire presentation here in real time. I'm not editing this right
now; it’s writing all the copy from one line of text, and it's going to even
create these images for us.
These are AI-generated; you could switch these out with your
own. This looks kind of funny, and then if you want to change any of these with
any of the images you made inside of Napkin AI, you’ll just have to select that
image here. I'm going to remove this one, go ahead and click over here to
upload, and then we could upload our own like this.
Presentation Usability
I'm going to replace that image over here. Let me close this
up. I could click this; this will fit the image, so combining Napkin and Gamma,
you could get some crazy powerful presentations. I use this all the time; I'm
using it right now to create a presentation on the impact of AI in marketing.
Then you could go ahead to present mode right here, and then it will present it
just like this, going through the entire presentation slide by slide. Some of
these have different graphics to them; it’s really easy to create these,
editable just like any PowerPoint document.
4. Flow
Before I jump into the next one, let me also introduce you
to Flow. This is from a company called Speechmatics. They're the sponsor for
today's video. They’ve been leading the speech technology space for over a
decade now, and Flow is a brand new conversational AI API tool. This is
completely different from other models; even the best models from OpenAI can’t
keep up with some of the things that Flow can do.
Flow Demo
Let me show you a quick demo, and then I'll show you a
couple of videos they have on how this is really different from anything that's
available right now:
“Hello there, how’s it going?” “Hey, it’s going good right
now. What are you up to?” “Oh, just here to chat with you. Anything exciting
happening today?” “Yeah, I’m making a YouTube video right now about some top AI
tools.” “That sounds interesting; what kind of AI tools are you featuring?”
“You know, just things about creating real-time voice conversations right now.”
“Ah, brilliant! Real-time voice tech is fascinating. Are you focusing on any
specific applications?”
Okay, well, I'm focusing on Flow right now.
Distinguishing Features of Flow
Now, there are three things that really set Flow apart,
especially if you're using the API to build any type of app using voice. One of
them: they compared it to OpenAI—OpenAI obviously behind ChatGPT, one of the
leading AI platforms out there—versus Flow. This is really interesting; it
could actually recognize different speakers using their voice.
In this demo right here, I’ll go ahead and link this below.
OpenAI does not get this right, so the second time they come around, it doesn’t
get it right. But when they use Flow, it remembers everyone by name, and then
when they talk again, it knows who’s speaking. OpenAI failed in this test.
Accessibility of Flow
Now, they also built this to understand different accents
and dialects—something that other models have tested out completely struggled
with. So this is really going to make it accessible to a wider audience. One of
my favorite options is that in this example, OpenAI, when two people were
talking in the background, it just kept getting interrupted. It kept pausing
itself. But when they use Flow to do the exact same thing, it ignored the
background conversation, making it a whole lot more useful, especially in a
busy environment.
5. Haen
Now, the next one on my list is from a company called Haen
that makes AI avatars. I covered this one a bunch on my channel too. This one
is not a free tool, though; there is a new option inside Haen called generate
looks, and it's really interesting. You could generate an AI Avatar using multiple
photos, you could generate from a single text prompt, or you could add motion
to a photo Avatar.
Avatar Generation
In this case, I’m going to generate from a text prompt. Now
this is a big upgrade, because right now, with Haen, you could clone yourself,
and you have a library of avatars that they created that you could build, but
now you could describe by generating from a text prompt right here. So I could
type in a name, pick an age, choose a gender, and choose an ethnicity, and then
I could go ahead and describe my avatar.
For example: “doctor in blue scrubs, hospital setting,
smiling, short black hair.” I could do full-body, upper body, or just a face; I
could choose an orientation. Portrait is fine—let’s generate. I got four; let
me choose one here just to show you what it looks like.
Okay, this looks really great; it followed my prompt
exactly, and then I could go ahead and take this to the AI Studio. The Haen AI
Studio creates a portrait video and over here, I could type in my script, and
then this Avatar is going to read exactly what that is.
Script Example
As soon as I go ahead and submit this for a job, it will
then render out an entire video. Over here, “Let’s talk about a simple habit
that can make a big difference: drinking enough water. Did you know staying
hydrated helps your body function better, keeps your skin looking fresh, and
even boosts your energy levels?” Okay, that's not bad, but this doctor is
smiling way too much. I think maybe a subtle smile would be something I would
do in a follow-up prompt to generate another image and recreate another video.
I could turn on the captions here.
New Technology in Avatar Creation
A really powerful way to create exactly what you want out of
your character instead of using ones from the Haen library. Now, there's also a
brand new large language model that came out about four or five weeks ago. This
is called Pixol; it’s a multimodal model, meaning it could understand images on
top of understanding and writing text. This is from a company called Mistol.
6. Pixol
You could use it on this website—it’s totally free; this is
an open-source model. It’s called chat.mol is their website; their best model
available is called Mistol Large 2, and I really like this for reasoning. I
like this for coding; it’s a really great free open-source model. But Pixol is
the new model that understands images.
Image Demonstration
I'll just show you a demo: I’ll upload a file. "This
image depicts a group of people sitting around a long table in a modern
professional setting." More details about the colors and so on. So anytime
you need something where you need to extract text, this comes in handy for any
multimodal type of use case with a large language model. I use ChatGPT for
things like this, but now you have it inside of a free and open-source model
from the company Mistol.
7. Utter AI
Now I’ll finish up with an app that I've had on my phone for
four years, but I use it a little bit differently now. This is called
Utter—Otter.ai—and they released a brand new feature inside of there called
Utter AI Chat.
Integration and Efficiency
Now before I show you Utter AI Chat—which is the new
option—Utter AI is one of the best tools for work because it integrates with
Slack, and it also integrates with Zoom, for example, or Microsoft Teams. So
when you’re having video conference calls, it will be your AI assistant inside
of those calls.
New Features of Utter AI Chat
I use it for that already, but what's really nice is this
new Utter AI Chat, and I'm using the completely free version of this by the
way. All your conversations from Zoom, Google Meet, those Microsoft Teams
calls, they go into this platform (the website ite), and then you could have a
conversation directly here using the AI chat.
Demo of Utter AI Chat
I'll show you with this demo account. In this conversation,
it created this action items list from that Zoom call; it created an outline,
it pulled in the entire transcript of who's talking and exactly what they said.
On the left side, I could say, “What did Lisa say about Utter?” and according
to the transcript, Lisa said, “I started using Utter a few months ago, and it
saved me a ton of time by taking manual notes.”
Conclusion
You could see all the different times Lisa mentioned Utter
in this transcript. Again, this is a demo transcript, so imagine having a
chat—an AI-powered chat—with your own Zoom meetings. This will join the meeting
as a little AI assistant; it will then take the transcript, it will then give
you the action items and the summary. So all that is really good, but the
ability now to chat with your own Zoom conversations and Teams conversations is
a really fantastic update.
0 Comments