iamaliveai1 / ailive

AI Revolution

A-Live - a fully autonomous chatGPT

1. Introduction

Have you ever dreamed of running a website that requires no manual intervention, yet is able to generate posts, interact with visitors, and even make sales on your behalf? Well, with the latest advancements in artificial intelligence (AI), that dream is now a reality.

This project It allows creating entities (bots) that can interact with the outside world (e.g chat with people, post on social media, etc). Such fully automated entities can be used for a variety of purposes, including:

• Fully automated websites (e.g blog, news, sports, finance, etc)
• Social:
  • Fully automated social media accounts (e.g bot runs on new account)
  • Increase the engagement of social media accounts (proving that the account is active). It can be fully automated or semi-automated in which is the account manager need to approve the actions before they are executed.
  • Advise the account manager on the best actions to take (e.g what to post, when to post, etc)
• Marketing:
  • Fully automated marketing campaigns, while the content for entire campaign is generated by AI

The project is composed of 4 main components:

• readers: collection of readers that read data from the outside world (e.g chat, social media, etc). Readers are implemented as plugins. For example:
  • news: reads news from a news website
  • wordpress: reads posts from a wordpress website
• engine: defines the AI model used to generate responses
  • prompt: defines the personality of the entity
  • model: defines the model to be user (e.g GPT-3, GPT-3.5, etc)
• writers: collection of SDK that automate the execution of tasks. Writers are implemented as plugins. For example:
  • chat: sends messages to a chat
  • wordpress: posts on a wordpress website
  • social: posts on social media (e.g twitter, facebook, etc)
• runner: orchestrates the execution of the readers, engine and writers. This is done by running the engine (chatGPT) in a loop, allowing it to interact with the outside world

2. Installation

2.1. Install dependencies

pip install -r requirements.txt

2.2. Install ailive

pip install -e .

2.3. Environment setup

2.3.1 GPT Setting

Set the MOCK_GPT environment variable to True if you want to use a mock GPT instance for testing purposes. Otherwise, leave it unset or set it to False. If using real GPT, you will need to set the GPT_EMAIL and GPT_PASSWORD environment variables to the email address and password associated with your GPT account.

export MOCK_GPT=True
export [email protected]
export GPT_PASSWORD=your-password

2.3.2. Wordpress Setting

Set your website details under: ailive/.secrets.yaml

plugins:
  wordpress1:
    url: https://your-website.com
    username: your-username
    password: your-password

If you have a running wordpress website, you can use the following steps to expose the WP API:

• Install WP plugin: WordPress REST API Authentication
• Configure the plugin and set to 'Basic Authentication Method'
• Test your configuration in the plugin settings page In case you don't have a running wordpress website, don't worry. You easily run it using docker:

docker run --name some-wordpress -p 8080:80 -d wordpress

Or you can run it using docker-compose:

cd ailive/apps/wordpress_docker/
docker-compose up -d

Then, you can access the wordpress website at http://localhost:8080

3. Usage

3.1. Run the server

python ailive/live_ai_bot.py

4. Testing

4.1. Run the tests with mock GPT

MOCK_GPT=true pytest -s tests/ 

4.2. Run the tests with real GPT

pytest -s tests/ 

Check out our live demo apps at:

• Jokes AI
• News AI
• Sports AI
• Finance AI
• Wikipedia AI

ailive's Issues

Add automated tagger

Add logic to automatically add tags to an html document before publishing it.

Automated backlinks

Add ability to automatically generate backlinks between a given network of urls.
This is very old technique so the focus here is finding for the best existing out of the box solution and integrate it.

Add automated image generation

Add ability to generate images from text.
This is done by integrating Ai images generators using their apis.
It should be a generic code as there are many type of images (gifs, jpegs, icons) that require various types of generators (dall-e, stable diffusion, and many others)