Loading video player...
****Note**** : In the video, one code line is missing from the Jupyter notebook which provides the columns and head info to the SYSTEM_PROMPT.. Here is the code for updating the system prompt SYSTEM_PROMPT = SYSTEM_PROMPT.format( columns=df.columns.tolist(), head=df.head().to_string() ) -------------------------------------------------*** Before thinking about building an AI agent, we need to understand what's an agent first and then how to build a simplest possible agent with the simplest possible available tools. one we do that, we're in a better position to decide and understand the usability and the need of an AI Agent. In this video I've create a simple AI agent using python Langchain framework which will write the python pandas query for me which will be running on top of the titanic dataset downloaded from kaggle I will ask my query in a human lanaguage and the AI agent will write the code in python pandas to share with me the results of the same. This video also explains the tooling or tool chain support with the LLM and AI agents.. Tools are the execution blocks or the programs which helps our AI agents to access, execute and return the result in data #aiagents #aiagent #llm Timecodes 00:00 What's an AI Agent 00:13 AI agents answering questions on Titanic Dataset 00:55 AI agents writing complex database queries 01:18 Creating an AI agent with Python Lang Chain 01:33 What's actually an AI Agent is? 02:27 The Brain and Purpose of an AI Agent 03:04 Creating an AI agent and Security Considerations 04:27 System Prompts and User Prompts 06:07 Tools and tool support in the LLM and AI Agent 07:03 Tool for executing parameter code 08:10 Integration LLM with AI Agent 08:50 Creating an AI agent with Python Langchain 09:47 Running the python langchain based AI Agents 10:20 Streaming Interaction between AI agent and LLM #theoryofcode #thetheoryofcode #theoryofproduct