phidata

An open-source platform for building, deploying, and monitoring AI agents with memory, knowledge, tools, and reasoning capabilities.

Phidata

We are going to create AI agents using Phidata. Phidata is one of the most popular tools which is open source and you are able to create AI agents, AI applications, and AI automation. In this video, we are going to create AI agents, give tools to those AI agents, add memory to those AI agents, and finally create a user interface. 

Even if you're an absolute beginner, you should be able to follow and create AI agents which can be used to solve complex tasks, to automate tasks, and automated decision making. You should be able to find all the code in the FI data repo under cookbook agent11, and I'll put the link in the description. 

So AI agents are nothing but independent AI systems working together to solve complex tasks. In this, we are going to build a web search agent and financial analyst agent, so they're going to work together to complete a complex task. We are also going to set up a user interface for these agents. The UI looks like this.

Where you are able to access those agents which you have created and able to give a task such as search about nmedia Tesla. Then also check their stock prices and give them this task. Then it's going to use tools such as web search tool and then Yago Finance tool work together and give us a final answer that is more relevant. If we try to do this manually, this will take time. But these agents speed up the process. 

This is a simple example, but you should be able to extend this from here. Building agents with memory, knowledge tools, and reasoning. Build teams of agents, chat with your agents, monitor those agents, and finally build an agentic system with an API database and Vector database. 

But before that, I regularly create videos in regards to Artificial Intelligence on my YouTube channel. So share the AI Agent Store with colleagues.

Like you so these are the steps which we are going to follow. First setting up the environment. Next importing packages. Next AI agents with tools and memory creation. Then creating the user interface. 

So step number one set up environment. First Step pip install open AI duck du go search and all these packages including Phidata and then click enter. You'll have all the code in the description below available in the cookbook step-by-step instruction. 

Next export your open AI API key like this and then click enter. All of these things you are doing in the terminal openai API key you can generate from platform.

Open.com we are going to use G pt40 that is a large language model. Now step number two importing packages. So I'm going to create a file called app.py and I'm going to open it. 

Inside the file First Step importing packages from fight.

Agent import agent then importing open AA chat then dougd go Yahoo finance tool SQL agent storage and serve playground app dugdug go is used for internet search. Yahoo finance is for getting stock price. SQL agent for storage and playground and serve playground app is for UI. 

Importing done now step number three creating AI agents with tools and memory. Create agents with tools and memory so the first agent is the web agent. So you just mention agent with the brackets that is a function providing the agent ID the role search the web for information the model name gp4. 

We are providing tool that is dugdug go search to search the internet. Then we are giving instruction always include sources and then we are storing the data with the table name and the file is going to be local. So it's going to save everything locally on your computer so your data remains private and mock down equals true. 

So that is the first agent so.

If you see this it consists of web search agent and financial analyst agent. So the web search agent will have web search tool that is Du go search and financial analyst agent will have yahu Finance tool. 

So next step is to create this financial analyst agent. So in the same page financial analyst agent same as before providing the agent ID role model tools as Yahoo finance and the instruction always use tables to display data. 

We are also storing the data in the same database and when you see here you need to create AI agent team. That means a team to make these agents work together. So that's when we have agent team. 

So here we are teaming up both of these agents web agent and finance agent with the storage. That's it. Now the final step is to create the user interface. 

So in the same code what we're going to do is set up UI setting up the playground with those three agents Finance agent web agent and agent team do.

Get app just to get the app. If name equals main, serve playground app. App playground and reload equals true. So app is the name of the file that is app.py. That's where we are creating it. And then playground, that's it. 

So as a quick overview, first we set up the environment. Then we imported the required packages. Then we created agents with tools and memory. And finally, we set up the user interface. 

Now I'm going to run this code in your terminal. Just type Phi and aut because this is going to authorize the user interface. So after this, click enter. Then you can log in and it will automatically authenticate it for you. As you can see here, CLI authentication succeeded. So you might need to use your username and password, and this logs into fi dat.

Apppp now in your terminal just type fyon app.py and then click enter. Now you can see here is the URL to the user interface and the application currently running locally. But the UI you can see in this URL. I'm opening that URL and here you can see the endpoint is Local Host. 

But here we can see all the agents Finance agent, web agent, and agent team. So even if I want to test let's say web agents and I'm going to ask what is the latest AI news, then it's going to use DD Go News tool search for relevant information and then it's going to provide us the information in this way. We are able to test these agents individually. 

But now I'm going to make it more complex by choosing the agent team and I'm going to ask search about Nvidia and Tesla. Also find the stock price going to click enter. So this is going to go step by step. First, it's going to use the web search agent and search about Nvidia and Tesla. Then it's going to

Use the finance analyst agent to get the stock price and here is the output as we know it. Used web agent and finance agent and this includes news about Nvidia and Tesla. 

And also it includes the current stock price that is super cool. Imagine the things you can build and test like this. I'm really excited about this and I'm planning to create more videos in regards to Phidata so stay tuned. 

Considering you already like Phidata, I also covered another Basics tutorial in regards to all the different tools, all the different functions, all the different types of agents in a beginners tutorial to Phidata. Which I'll put that link here, which I highly recommend for you to watch. Just click that and I will see you there.

Note, that there may be slight differences in text and video. Watch video for the original content.

Phidata Video