Transcription
Google is leveling up with the release of Gemini 2.5 Pro, a model specifically designed for advanced Agentic AI applications. It's marketed as the powerful tool for building intelligent agents and automating end-to-end tasks with precision and speed. More importantly, it's currently the most intelligent AI model available that's delivering top-tier performance while remaining cost-efficient. Beyond that, Gemini 2.5 Pro offers powerful multimodal capabilities, including image generation, text-to-speech, and advanced tool use. It supports structured function calling, which integrates search as a tool and can even execute code, making it an ideal complex agent workflow with top-tier performance in reasoning, coding, and multi-step task execution. It stands out as one of the most capable models available today.
Even more impressively, this Gemini 2.5 Pro model empowers you to build AI agents with built-in agent capabilities, completely for free, lightweight, and it's something that is extremely fast and intuitive, designed for speed and usability. They make sophisticated agent creation easier and more accessible than ever before, and that's exactly what I'll be showcasing throughout today's video. As I'll be using Vector Shift to help you create AI agents with powerful automation capabilities powered by the Gemini 2.5 Pro model.
For those who are unfamiliar, Vector Shift is an all-in-one AI platform that lets you build AI apps and automations through a simple drag-and-drop interface. It's incredibly easy to get started, and best of all, it's completely free. So let's dive straight to it. To get started, head over to the link in the description below and click on "Get Started." Once you have done so, you want to go ahead and create an account if you do not have one already, or if you do, simply go ahead and log in. Once you sign in or sign up, you're going to be then greeted to the main dashboard of Vector Shift, and this is actually a new revamped look which looks absolutely amazing. But essentially, this is where you can manage all your pipelines, all the different templates that you can actually adopt, meaning that there are ready-made templates that you can easily work with, which will help you automate many of these different types of integrations as well as different categories, from knowledge assistance to chatbots. You can actually take a look at all these pipelines, knowledge files, further in detail on the left-hand panel.
Now what we're going to do is take a look at creating our own pipeline from scratch, cuz we're going to be showcasing the capability of creating AI agents quite easily with Vector Shift and then having Gemini 2.5 Pro powering them. Before we get started, I just want to mention that you should definitely go ahead and subscribe to the World of AI newsletter. I'm constantly posting different newsletters on a weekly basis, so this is where you can easily get up-to-date knowledge about what is happening in the AI space. So definitely go ahead and subscribe, as this is completely for free. What we're going to be doing is automating the process of handling invoices directly from our email. We're going to be extracting key details like the price, invoice number, and other important fields, all without writing a single line of code. So let's get started.
This is the drag-and-drop interface of Vector Shift that will let you configure your AI agents with different nodes. These nodes are a way for you to build your AI agent. What we're going to be doing for our particular case is placing down a Gmail node, which is where it will trigger the pipeline to run our agents based off of the invoice that comes inwards to the Gmail. Now what I'm going to be doing is showcasing this example so you get a better idea as to how you can build your own AI agents powered by Gemini 2.5 Pro within Vector Shift. So what we'll do now is go ahead and configure this Gmail trigger node. What we want to do is select the trigger event, which is where whenever emails are sent within our main inbox, it will trigger this node. So let's go ahead and then connect our account, which we have it as this World of Crypto Gmail account, and then what we're going to do is select the trigger. This is where we can select different triggers, so whenever a new mail is sent within the inbox, we want it to then process this trigger node. So let's go ahead and confirm this.
Now say if we have invoices being sent to our email, but say that there are multiple invoices, how do we actually manage that? Well, this is with the list node feature. The list node feature is something that will provide consistency when processing multiple attachments within our email. Now this is where we're going to need to actually create a new pipeline, and the reason why is because we need to process the extraction of whatever we're dealing with within this pipeline so that it could then trigger what will happen in the second pipeline, which is the master pipeline. So for this particular pipeline, we're going to be extracting the contents from the invoice by placing down an input node and then going over to the data transformation tab and placing down an AI operation node so that we can extract the contents and then maybe have it so that whatever contents we extract could be displayed or outputted to a Google sheet. So what we want to do is give this input node a name, and we want it to actually process files because this is essentially where it's going to process the invoice once it triggers the actual email. Now the output fields, we want it to process text, and we want to have the AI operation extract the data from the invoice. So click on "Extract Data," and then we can go ahead and connect these two nodes. Now what we want to do is drag the invoice to the text for extraction and then process the text from it. Then we need to go ahead and provide a couple of fields of things we want to extract. So, for example, we can extract the vendor name; we can also extract the date. So we can then go ahead and provide a date. Then I have provided a description of what I want the format to be outputted as. Now, as we all know, there's various types of invoices, so what we can also do is place down another AI operation node, and what we can do is select the categorization feature. Essentially, this is where we can then connect the input node to this categorization node, which will help us categorize different types of invoices, from food to utilities, etc. So now I have solidified all the different category names and give it a category description to process the different fields and to categorize them.
Now what I want to do is, now that we have the extraction process set up, I want to have it outputted to this Google sheet. So you may ask how do I do this? Well, it's actually pretty simple. You head over to the integration tab and you want to place down a Google sheet node, and what you want to do is have it so that it adds a new row. So we're going to select our email; make sure that you have this fully connected, then you can go ahead and click "Next." You want to then select a file that you want to have for the new rows to be added, and then you want to click on "View Folder" and then select the spreadsheet that you've just created, and you can see the three columns that we have within our sheet are now displayed within this pipeline. Next, you just need to configure the input fields for these two different nodes. For example, the extraction data will be outputted as vendor and invoice data, cuz those are the two fields we have written over here. So what you can do is press on this plus sign, and then you can go ahead and click on the extract data node, and you can then select vendor for this task. Now for the second one, you want to then do the exact same thing; click "Extract Data," but you want to have it so that it selects the invoice date, cuz that's what's going to be extracted and then outputted to this particular info field. In the same manner, you can do that for the categorization node. So click on the plus sign, then click on "Categorizer," and you can see that now you can process all these different separate nodes and then input the fields within our Google sheet.
Now this next step is optional, but what I'm going to be doing is, now that we have all this information extracted, I wanted to actually send a summary back to the email so that it could provide a good idea of what had happened. To do this, it's super simple. Go to the general tab and place down a text node, and you want to specify the different text categories that you want to process with natural language. Then you want to go over to large language models and then you want to place down a Google large language model node. You want to select the Gemini 2.5 Pro, and then what you want to do is give it a system instruction. This is where you can specify what you want to have processed. So what you can do is go ahead and connect these two models. You want to then process it as text, and then what you can do is output it, and there we go. We now have this pipeline fully configured. We're not done, but in summary, we start off with the email being sent in, and then the second pipeline will process the file with the node list feature, and then it will extract the data, categorize it, and then add it to the Google sheet that we have created. But then what it will do is take another step to then process the information, summarize it, and then send an email back, which will talk about what had happened. So now that we have everything configured, go ahead and click "Deploy Changes" and then click "Deploy." Now go back to your master pipeline, click on "General," and what you can do is place down a pipeline node. This is where we're going to connect the previous pipeline that we had created. So go ahead and find that pipeline that we had. So I have now added the pipeline, and then I'm going to turn on list mode so that it can process multiple files. Now what I've done is placed down a data transformation node as list ops. This is where it's going to go ahead and have it flatten the list of lists because you want to have it processed as real text. So make sure you go ahead and place this node down. Then what you can do is you can then select the list to join, which is by simply connecting the node, and then I've placed down another Gmail node so that it can go ahead and give another reply back, so it will send a reply back to the person who sent the email, and then it will process this flattened text, and then the email ID. We have this fully configured, and what you can do is just simply click "Deploy Changes."
So let's test this out. I'm going to go ahead and send an invoice, and what we're going to do is trigger this whole pipeline and then output the text and then have it send another email back to the World of Crypto email, and there we go. Just like that, we have an output of the summary of the two invoices I had provided, and then we also have it inputted through our Google Sheets. This is where we have this full pipeline of processing emails fully automated, thanks to the Gemini 2.5 Pro and Vector Shift. You saw how easy it was, and it took me a matter of a couple of minutes to fully set this up, and this was all capable thanks to Vector Shift. So with that thought, guys, I hope you enjoyed today's video and got some sort of idea as to how you can set this up. This is probably the easiest way for anyone to build AI agents, especially powerful ones like the Gemini 2.5 Pro agents. So with that thought, guys, I hope you enjoyed today's video, got some sort of value out of it. I'll leave all these links in the description below. Subscribe to the new channel, follow me on the newsletter, join our Discord, follow me on Twitter, and lastly, make sure you guys subscribe to the YouTube channel, like this video, and please take a look at our previous videos cuz there's a lot of content that you will truly benefit from. But with that thought, guys, have an amazing day, spread positivity, and I'll see you guys fairly shortly. Peace off.