STORM AI: Create Full Wikipedia Pages in Minutes With AI Agents (Tutorial)

Introduction

A new intelligent project out of Stanford, known as STORM, allows users to create entire Wikipedia pages on various topics through the combined effort of multiple AI agents. This innovative tool can run locally on your computer, ensuring efficient and localized processing. Remarkably, STORM not only compiles comprehensive Wikipedia pages but also includes references to the sources of its information.

Introduction to STORM

STORM stands for Synthesis of Topic Outlines through Retrieval and Multiperspective Question Asking. This project uses advanced AI and web search capabilities to gather information from multiple sources and generate structured Wikipedia-like articles. The advantage of STORM is its precision in citing references, which enhances the credibility and traceability of the information provided.

Installation Guide

To get started with STORM, you can follow the setup process described below. The project is open-source, available on GitHub, and can be run locally using the OpenAI API for processing.

Prerequisites

VSS Code: This is the text editor we will use to run and manage the STORM code.
Python 3.11: Ensure you have this version of Python installed on your system.

Step-by-Step Installation

Clone the Repository: Open VSS Code. Navigate to the directory where you want to store STORM (e.g., the desktop):
```
cd desktop
```
Clone the repository:
```
git clone https://github.com/your/repository/storm.git
```
Create and Activate Python Environment: Create a new conda environment:
```
conda create --name storm python=3.11
```
Activate the environment:
```
conda activate storm
```
Install Required Packages: Install dependencies from the requirements file:
```
pip install -r requirements.txt
```
Add Secrets File: Create a secrets.toml file in the project’s root directory:
```
touch secrets.toml
```
Open the file and add the necessary API keys:
```
[openai]
apikey = "your-openai-api-key"

[u.com]
searchapikey = "your-u.com-api-key"
```
Run STORM using Streamlit: Navigate to the demo_light directory and launch the Streamlit server:
```
cd demo_light
streamlit run storm.py
```
Using STORM: Open your web browser and go to localhost. Enter a topic and click on research. STORM will perform comprehensive research on the topic and display an organized article with references.

Features

Human-AI Collaboration: A planned feature for collaborative generation of content.
Local Running: Besides web scraping, all other functions can be locally executed.
Support for Local Models: There's support for local models like VLM with Mistool and PLLMS.

Keyword

STORM
AI Agents
Wikipedia Generation
Local Running
OpenAI API
Comprehensive Research
Human-AI Collaboration

FAQ

What is STORM? STORM stands for Synthesis of Topic Outlines through Retrieval and Multiperspective Question Asking. It is an AI-driven project that generates well-referenced Wikipedia-like articles on any topic.
Can STORM be run locally? Yes, STORM can run locally on your computer using the OpenAI API, except for the web search component. The rest of the processing can be done locally.
What do I need to get started with STORM? You need VSS Code, Python 3.11, an OpenAI API key, and a u.com search API key.
How do I add secrets to the STORM setup? Create a file named secrets.toml in the root directory and add your OpenAI and u.com API keys.
Can I use STORM without an OpenAI API key? Currently, STORM requires an OpenAI API key for its core functionalities, although there are efforts to support local models.
How does STORM ensure the information is credible? STORM provides references to each of the sources from which it collects the information, making it easy to verify the accuracy of the generated content.

If you found this tutorial helpful, please consider liking, subscribing, and letting us know if you want a full local tutorial on how to use STORM without relying on external APIs.