Open-source tool to seamlessly connect multiple AI models in repeatable flow.
While a live demo is available for convenience, for the best experience we recommend running the application directly on your local machine.
🎉 v0.4.0 is Now Available 🎉
🌟 UI Enhancements |
AI Flow is an open source, user-friendly UI application that empowers you to seamlessly connect multiple AI models together, specifically leveraging the capabilities of ChatGPT. This unique tool paves the way to creating interactive networks of different AI models, fostering a multi-perspective response to various prompts.
Imagine setting up a simple workflow where:
- One node generates a poem
- One node gives a review of the poem
- One node creates a prompt for the cover of the poem
- One node makes the cover with Stable Diffusion
If we want to change the idea of the poem, we can edit the first node, and keep the same flow.
This flow logic can then be applied to different inputs, yielding varied results for similar subject matter.
AI Flow is particularly useful for content creation. By enabling different GPT models to adopt different roles, AI Flow orchestrates the creation and evaluation of digital content, from social media threads to blog articles.
Our primary goal with AI Flow is to democratize access to AI interactions, enabling anyone to create and manage their own AI networks right from their personal computer. In addition, we value your autonomy and control over costs. With AI Flow, you utilize your own OpenAI keys, giving you full authority over your budget. Importantly, the software runs locally on your machine, assuring you of privacy and control over your data and operations.
Features
In a nutshell, AI Flow provides a visual platform for crafting and managing AI-driven workflows, thereby facilitating diverse and dynamic AI interactions.
- 🎨 It offers a drag-and-drop interface to design these workflows
- 📊 Monitors their execution in real-time
- 🗂️ AI models can be conveniently managed and organized
- 💾 Workflows can be exported or imported for sharing or backup purposes
Other basic use cases
Summarize web pages & youtube transcript
Use pre-made prompts to build quickly
Split input with AI
By combining those features we can create different flows.
Installation (Windows executable)
For a quick local setup, grab the .exe from the repository's releases section.
Installation (without Docker)
Prerequisites
Before getting started, make sure you have the following dependencies installed on your system:
Clone the Repository
- Clone the repository:
git clone https://github.com/DahnM20/ai-flow.git
- Change to the project directory:
cd ai-flow
UI Dependencies
- Go to the UI directory:
cd packages/ui
- Install dependencies:
npm install
Backend Dependencies
- Go to the backend directory:
cd packages/backend
- Install Python dependencies:
poetry install
For Windows only
- Launch poetry shell :
poetry shell
- Install the windows requirements in the poetry shell :
pip install -r requirements_windows.txt
Usage
- Start the backend server:
cd backend && poetry run python server.py
- Start the ui application:
cd ui && npm start
- Open your browser and navigate to
http://localhost:3000
- Use the drag-and-drop interface to design your AI workflow
- Connect AI models and define data flow between them
- Click "Run" to execute the AI workflow
- Monitor the execution progress and results in real-time
🐳 Docker
Docker Compose
- Go to the docker directory:
cd ./docker
- Update the .yml if needed for the PORTS
- Launch
docker-compose up
ordocker-compose up -d
- Open your browser and navigate to
http://localhost:80
- Use
docker-compose stop
when you want to stop the app.
Contributing
Contributions are welcome! If you find any issues or have suggestions for improvement, please open an issue or submit a pull request.
License
This project is licensed under the MIT License. See the LICENSE file for more details.