Project Setup Instructions

Prerequisites

Python 3.10 or greater: Download Python

Clone the Repository

Clone the repository to a directory of your choosing:

HTTPS:

git clone https://github.com/codebrain001/pycon-2024.git

SSH:

git clone git@github.com:codebrain001/pycon-2024.git

GitHub CLI:

gh repo clone codebrain001/pycon-2024

Obtain API Keys

Get the following API keys for the project:

1. OpenAI API Key

OpenAI models are the LLMs used for this project.
Obtain your API key: Where do I find my OpenAI API key?
Note: Requires payment or use the test API with limited credit for demo purposes.

2. Serper Dev API Key

For Google search functionality (first 2,500 search queries are free).
Get your API key: Get Serper Dev API key

3. Composio API Key

Utilized for Composio tools in the project.
Obtain your API key: Get Composio API Key

4. Perplexity AI API Key (Optional)

Perplexity is a free AI-powered answer engine that provides accurate, trusted, and real-time answers to any question.
For more advanced searching.
It is paid to get and use its API.
Get your API key: Getting Started with Perplexity API

5. Langtrace API Key

Langtrace is an open-source, external tool that helps you set up observability and evaluations for Large Language Models (LLMs), LLM frameworks, and Vector Databases.
Obtain your API key: Langtrace Quickstart Guide

Setup Virtual Environment

With the repository cloned on your system, access the pycon-2024 directory, which has the following structure:

├── Application-welcome-ui.png
├── LICENSE
├── Meeting-notes-001.pdf
├── README.md
├── req_engineering_project
├── requirements.txt

Create and activate a Python Virtual Environment

macOS and Linux

python3 -m venv venv
source venv/bin/activate

Windows

python -m venv venv
venv\Scripts\activate

Install Crewai packages

Install the packages listed in requirements.txt:

pip install -r requirements.txt

Setup CrewAI Project

With the CrewAI packages successfully installed, navigate via the terminal to the req_engineering_project folder:

cd req_engineering_project

Directory structure:

.
├── README.md
├── poetry.lock
├── pyproject.toml
└── src

Note: When you create a new CrewAI project, it uses Poetry to handle packages and dependencies in a virtual environment more efficiently, avoiding dependency conflicts.

Install Packages with Poetry

Poetry is already installed when we installed the crewal packages. Next, install the project dependencies using Poetry:

poetry install

Configure API Keys

Upon completion, input your API keys in src/req_engineering_project/tools/search_toolset.py

import os

# Load environment variables
os.environ["SERPER_API_KEY"] = "your-serper-api-key"
os.environ["COMPOSIO_API_KEY"] = "your-composio-api-key"

Optional: Add Perplexity AI Integration If you want to utilize the Perplexity AI search tool, add the integration connection to your account as follows:

composio add perplexityai

You will then be prompted to input your Perplexity AI API key during the setup process.

Update the Langtrace API key in src/req_engineering_project/app.py:

LANGTRACE_API_KEY = "your-langtrace-api-key"

You're all set!

Running the strealit application

You can run the streamlit application via:

streamlit run src/req_engineering_project/app.py

For this project, we will be utilizing a simulated meeting, preview here. Feel free to create a simulated meeting for a different use case to test the application.

The Application Home UI is seen below

Workshop Highlights

🚀 Using the CrewAI latest workflow: crewai==0.60.0
🛠️ Utilizing tools from Langchain, LlamaIndex, Composio
📊 Sequential and Hierarchical process workflow
📝 Agent Planner for step-by-step plan before execution
🔍 Utilizing Langtrace for monitoring and observations of LLMs, LLM frameworks, and Vector Databases (Chroma DB)
🤖 Utilizing OpenAI models via APIs: GPT-4o, GPT-4o-Mini, and o1-Mini
🦾 Running agents with same LLM or different LLMs (via LiteLLM)