Skip to content

Latest commit

 

History

History
121 lines (91 loc) · 3.61 KB

README.md

File metadata and controls

121 lines (91 loc) · 3.61 KB

🤖 AI Workflows Framework

Transform AI Assistant Chats into Powerful Python Code

Why This Framework?

You love working with AI assistants - writing instructions in plain text, having natural conversations, getting quick results. But sometimes you need more:

  • Want to run the same workflow reliably, again and again?
  • Need to process data from multiple sources first?
  • Want precise control over how the AI thinks?
  • Need to integrate with your existing tools?

That's where this framework comes in. It helps you transform your text-based AI workflows into Python code that:

  • Runs independently of any AI assistant
  • Uses proven open-source libraries
  • Gives you complete control over execution
  • Adapts to your needs

Prerequisites

  1. Python Environment

    • Python 3.10 or higher
    • Poetry (package manager)

  2. API Access to AI Models

    • You can for example create a free account at Groq and generate an API key

Quick Example

See the FAQ Analysis example for a complete working example.

Installation

  1. Get the Code

    • Clone the repository: git clone https://github.com/cbardyn/ai-swiss-workflows
    • Navigate to package: cd _ai_workflows_packages/ai_workflows

  2. Install Dependencies

    • Install with Poetry: poetry install
    • Activate environment:
      • Linux/macOS: source .venv/bin/activate
      • Windows: .\.venv\Scripts\activate

  3. Verify Installation

    • Try running an example: cd ../../_ai_workflows/code_based/faq_analysis/_code && python run.py

Step-by-Step Usage Guide

1. Create Your Project Structure

Create three main files in your project directory:

  • config.yaml (AI configuration)
  • instructions.yaml (Your workflow tasks)
  • run.py (Python runner)

2. Write Your Instructions File

In instructions.yaml, define:

  • Your workflow goal
  • Tasks to perform
  • Input files to analyze
  • Output files to generate

Pro tip: Start with a text-based workflow in instructions_v1.0.0.md and use your AI assistant to convert it to YAML format.

3. Configure AI Tools

In config.yaml, specify:

  • Your AI model choice (e.g., Groq's LLaMA)
  • Your API key
  • Model parameters (temperature, etc.)

4. Create Your Runner

In run.py, use the framework to:

  • Set up logging
  • Load your configuration
  • Execute your workflow

5. Run Your Workflow

Simply execute: python run.py

Getting Help

  1. Check the FAQ Analysis example for a complete working example
  2. Review the code-based workflows tutorial

Framework Features

Smart Context Management

The framework automatically:

  • Loads files and URLs
  • Creates searchable vector databases
  • Finds relevant information for each task

Task Orchestration

The framework handles:

  • Task sequencing
  • Information passing between tasks
  • Progress tracking

Error Handling

Built-in handling for:

  • Missing files
  • API failures
  • Invalid configurations

Possible Improvements

  • More connectors to external systems (Dropbox, Google Drive, etc.)
  • Better testing: No automated tests yet to ensure everything works perfectly
  • Better security: API keys are stored in simple text files
  • Better performance: Tasks run one after another instead of in parallel
  • Better reliability: No backup AI models if the main one fails
  • Better data safety: No automatic backups of your data
  • Better monitoring: Can't track how well the system performs
  • And more!
Built with ❤️ by AI Swiss