Transcription
Google Gemini just released their brand new AI agent builder that allows you to build out agents and automations directly with Gemini. And by the end of this video, you're going to know exactly what it is, how to use it, and I'm going to compare this to Chat GPT's new agent builder to see which one's actually right for you.
Okay, so in order to actually access this, what you're going to have to do is come to business.google.com. You're going to load this up. Now, I get tons of questions about this, so I do want to answer this. In order to actually use this, you're going to have to be on Gemini Business or Gemini Enterprise. Now, Gemini Business is $21 per seat. So, if you have multiple people on your team, it's going to be 21 bucks for each of them. If you're using Gemini Enterprise, it's going to be $30 per seat.
Now, this is what this actually gives you access to. Over here, we can see that this is brand new. So, there are several things that you could do here. First and foremost, I wanted to come up here and show you the different models that you can access. So, there's an auto model, there's a Gemini 2.5 flash model, and there's a Gemini 2.5 Pro model. I'm kind of disappointed that they only give you three, but if you click right here, you'll be able to see that you can still upload files, add to drive, add from one business. And if you click right here, you could have this search internal sources, which I'm going to show you in just a second. Search on Google. You can generate videos in here. You could generate images, or you can run deep research.
So, this is kind of similar to regular Gemini. Instead though, if you click on this right here, you will be able to manage all of your different connectors. So, you have your calendar, you have Confluence, you have Drive, Gmail, Gyra, One Drive, Outlook, SharePoint, and that's it. Those are the only connectors that this currently has. This doesn't have MCP access. It doesn't have additional tool access. Doesn't have anything like that.
Now, in addition to that, you can come over here. You could change what the appearance actually looks like. I would strongly suggest that if you're doing this, you have both for you and prompts on. As you can see, if we come over here, this actually walks through a bunch of sample prompts that we could actually do with all the different tools and different data that we gave this access to. In addition to that, this gives me for you right here. So, this actually tells me like what things are on my calendar and then different things that were uploaded to my drive and different things like that.
Now, in addition to that, what you're able to do with this, which is really why I think most of you guys are watching this video, is if you come over here to agents, you can build out agents. Now, there are two types here. There's one agents that you actually built out, and there's two agents by Google. Now, right now, they only have one Google agent, and it's just deep research, which is in my opinion very disappointing. But if we wanted to actually create an agent, we could come over here, click on create agent, and we can see that they have an agent designer right here. So, you literally tell this what you want, and then it goes off and does it. For example, create a Q&A agent trained on a set of documents. We're going to click on this right here. We're going to click submit on here. And then what we will be able to do here is this will actually build out the flow for this. It's going to go through and understand the goal. And then it is actually going to build out what that agent is going to look like for us.
Now I'm going to come over here and I'm going to load this up again in a new window while this goes through and actually creates this because kind of boring to sit here and watch this or actually it's already done. We could see right here what the actual flow looks like of the agents. If we click into this, we can see the name, we can see the model it uses, the description, the instructions here. We can see all the different data and tools that it has access to. We can see the details of this. So again, name, description, any knowledge. So we could train this on any knowledge if we wanted to. We could add in files here. We could add in different personalization things that we can actually use to set the tone of the conversation. And then we could preview this right here. These are those personalization things. And then as we could see, you could manage what data this actually has access to. And then this goes through and does things. I'm going to say, for example, what is the best time to post on Tik Tok, the best hashtags to use, and how does the algorithm work? Now, the reason that I'm asking this this is because I have a bunch of documents inside of my drive, which this should have access to. It does that will actually give this information. So, let's see if this is actually able to do this.
Now, we can see this is just the flow details of preview. We could come over here and we could create this. And we can see right here that this went through and actually look this stuff up. Did the same exact thing with the hashtags right here. Again, this is pretty cool that it's able to do this, especially if you're enterprise. In addition to that, it also did a Google search here, which is cool. But a few things that I don't actually like about this is one, as we could see over here, if we come into this and we ask this to create the same exact thing, it might actually create it differently. So, I had this create a meeting assistant here that will actually prepare me for my meetings by gathering and summarizing relevant information. And one thing that I found if we come over here and click into agent designer, but real quick, hold that thought because before I compare Gemini's agent builder to OpenAI's agent builder and I show you what I found inside of this, I wanted to make sure that you're not one of those people that's scared that they're going to lose their job because of AI because a recent poll just came out and it showed that more than 71% of people think that their job is going to be permanently replaced. But I want to help you not have that happen to you. Which is why I wanted to tell you that Outskll, the world's first AI focused education platform, is hosting a two-day mastermind workshop, which is happening this weekend on Saturday and Sunday from 10:00 a.m. to 7:00 p.m. Eastern on both days. And the best part is it's 100% free for you to attend if you're watching this video. Over 10 million people across the globe have already attended this training, and they've boosted their career growth as a result. And some people have even started businesses that are now making them thousands of dollars a month after they've learned about new AI tools. The training, which is rated 4.9 out of five on Trustpilot and usually costs over $395, is going to be completely free for you if you go to the pin comment below right now and sign up. But here's why I really urge you to join. Firstly, this isn't going to be just theory. This is going to be practical things that you can actually implement from people that work at companies like Microsoft, Nvidia, and so much more. Secondly, in just 2 days, here's what you're going to learn. You're going to learn over 10 plus AI powered tools. You're going to learn how to use AI and Excel Sheets and presentations. You're going to learn how to build AI agents and automate your workflow and so much more. In addition to that, you're going to get access to the learning dashboard where you're introduced to like-minded individuals to help build with them and enhance your overall learning experience. And that's not all because you're going to be walking away with more than $5,000 in bonuses. You're going to get a prompt bible, a road map to monetize AI, and a personalized AI toolkit builder that's only available if you attend both days. And both me and my entire team are going to be participating in this. So, please make sure that you go to the pin comment below and you get signed up, too. PS, make sure you join the WhatsApp community in order to stay up to date on exactly when the trainings are going to begin.
Is that one time that I designed it, it gave me a completely different flow. So, this one right here accesses all three things at once. it had another one that had another flow. So if we click on here, we could see that we could have a main agent and then we can actually create a sub aent and then we can create other sub aents off of this if we wanted to. So if we come back over here, we're going to get this plus again. We can actually get this to create multiple sub aents in order to handle multiple different tasks.
Now one other thing that we do need to cover here cuz as you can see this is somewhat limiting in my opinion. If we come over here, they basically partnered up with a bunch of different companies. If we come over here, and I'm going to do a search for partners, we should be able to find it pretty easily over here. Let's see. Okay, so if we come over here, AI agent finder. If we click into this, this is where you can actually find different agents that use different tools. For example, Box built out a specific agent or there's an airline assistant or there's an academic and skill navigator from Asenture. We could see that there are a bunch of different things. So, let's say that you needed something to help with cash flow analysis. You'd be able to come over here and ask. This would then search through the agents and say, "Hey, you should use this one." But, as we see here, if we come over here into view purchase options, if we come over here, we can see an overview of this. We can see pricing for this, but it doesn't actually tell us anything. We actually have to contact the vendor in order to get this agent put inside of our Gemini Enterprise.
Okay. Now, in terms of actually comparing Gemini Agent Builder to OpenAI agent Builder, I want to make sure that I'm fair here because they both have things that I like and they both have things that I dislike. So, let's start with Gemini over here. The one thing that I do like is that they actually have this agent designer. So, you tell it exactly what you want it to create and it goes off and it actually creates it. I think that that is really good. I think it's really useful. I think it's great for adoption and I think it's going to get a lot of people to build ad agents quicker.
Now, what I don't like about this is the connections that they have are incredibly limited here. I mean, there's only nine of them and they're all Google or related to Google or slightly off of Google in the sense that it's just like something like SharePoint, Outlook, One Drive or Gyra. Again, I think that this is incredibly limiting. I think that this kind of limits the use cases that companies can do if they just want to sign up for this, get into this immediately, and begin actually taking action and building things out.
Now, I do to some degree like the fact that they have this agent finder. I think that this is cool. Again, I think that this is a nice touch, but it's a pain that we have to then contact all these people individually. Again, I get the whole point that like this is enterprise level and that people are going to be paying six to seven figures in order to actually access these, but is that really the world that we're headed into? I don't know. I don't necessarily believe that. Um, I think that what AI really opens up is the ability for anybody to create anything and attach anything and be able to do anything. And I think a lot of the security concerns and things like that are going to be fixed much easier instead of like getting people to pay all this money in order to do these very simple things that otherwise you'd be able to do really easily. For example, Box AI agent for Gemini Enterprise is an AI agent that securely searches, analyzes, generates insights from your content within Box using Gemini Enterprise. You could build this out in any AI agent builder in like the next 30 seconds.
Now, I do think that OpenAI also has some pretty great things here. If we come over into OpenAI, we see that we could very easily fine-tune things. We can very easy evaluate things, which I think is pretty awesome. In fact, you could like upload like mass evaluations here to run them all at once, which I think is great. In addition to that, if we come over here into agent builder and we click on create, you will be able to see that there are a bunch of things that we could set up here. We could set up logic. We could set up different things with data. We could set up MCP servers. In fact, they have way more connections than Google does. We could see all of them right here. Or we could add in a custom one if we wanted to. For example, you could hook this up to rub.app. You could hook this up to Zapier and it will then give you access to all the tools that you could use. You could add in guardrails here. You could add in notes very easily. You could begin the agent, you could end the agent, you could do file search. And the other thing that's really nice about this is that you can then deploy this on a website or deploy it wherever you want. In addition to that, it gives you access to multiple different assistants, multiple different types of things like image generation, audio, transcription, text to speech. This gives you access to way more things in my opinion. Again, I think Google has some cool features, has some cool quirks, but in my opinion, Agent Builder is just one a lot easier to use and two allows you to do way more things and feels a lot less limited to me.
If you enjoyed this video, I would check out this video right here that walks you through several other changes that Gemini just released because it's not only this one. They launched five other ones that impact Notebook OM and a bunch of their other tools. I'll see you over there.