​
References and Acknoledgements

​
Features

  • Specialized Research: Dedicated researcher agent focused on discovering groundbreaking technologies
  • Intelligent Analysis: Powered by Meta-Llama-3.1-70B-Instruct model for deep insights
  • Structured Output: Well-defined tasks with clear expected outputs
  • Sequential Processing: Organized task execution for optimal results
  • Customizable Crew: Easy to extend with additional agents and tasks

​
Prerequisites

  • Nebius API key (get it from Nebius AI Studio)
  • If running locally, python 3.10 or higher dev environment.

​
Tech Stack

  • CrewAI agent framework
  • Nebius AI for LLM inference

​
Task Structure

Tasks are defined with:
  • Clear description
  • Expected output format
  • Assigned agent
  • Sequential processing

​
Example Tasks

  • “Identify the next big trend in AI”
  • “Analyze emerging technologies in quantum computing”
  • “Research breakthroughs in sustainable tech”
  • “Investigate future of human-AI collaboration”
  • “Explore cutting-edge developments in robotics”

​
Setup

The code can be run locally or on Google colab. Colab is recommended, as it doesn’t need any setup.

​
Local env setup

  1. Clone the repository: 
    git clone   https://github.com/nebius/ai-studio-cookbook/
cd  agents/crewai-research-agent
  2. Install dependencies:
    if using uv package manager     
    uv sync
    If using conda/pip 
    pip  install  -r requirements.txt
  3. Create a .env file in the project root and add your Nebius API key: 
    NEBIUS_API_KEY=your_api_key_here

​
Video

Video tutorial

​
Code

agent.ipynb notebook can be run locally or Google Colab. agent.py python script can be run locally. 
# using uv
uv run python agent.py

# or 
python agent.py

​
Dev Notes

How the uv project was created. 
uv init .
uv add -r requirements.txt
uv add --dev ipykernel   # for jupyter kernel
uv sync

# create a kernel to use uv env within vscode
source  .venv/bin/activate
uv run python -m ipykernel install --user --name=$(basename $(pwd)) --display-name "$(basename $(pwd))"
jupyter kernelspec list  # verify kernel is successfully created