Agentic Workflow

App Overview

• App Name: Agentic Workflow
• App Description: A flexible AI-powered text processing system that enables various language-based workflows using multiple AI models.

Core Functionality:

Features:

• Multi-Model Support: Integration with multiple AI providers (OpenAI, Mistral, Google, Anthropic)
• Workflow System: Predefined workflows for common text processing tasks
• File Management: Automated output handling and logging system
• API Integration: Unified interface for different AI model providers

User Actions (Stories)

As a user I want to:

• Process text through different AI workflows
• Choose between different AI models for processing
• Save and track processing results
• Access different text processing functions (translation, summarization)

App Layout and UI Elements

Main Interface Screen:

• Purpose: Text input and workflow selection
• Content: Input form, workflow selector
• UI Elements: Text area, workflow dropdown, submit button
• Navigation: Simple single-page interface

Available Workflows:

• Translation (CS-EN):
  • Translates between Czech and English
  • Uses Gemini 1.5 Flash model
• Text Summarization:
  • Creates concise summaries of input text
  • Uses GPT-4 Mini model

Technical Specifications

• Python/Flask backend
• Environment-based configuration
• Supported AI Models:
  • GPT-4 Mini (OpenAI)
  • Mistral Small
  • Gemini 1.5 Flash
  • Claude 3 Haiku
• File-based output system
• Comprehensive error handling

Future Enhancements

• Additional Workflows: Expand available text processing options
• Model Selection: Allow users to choose specific models for each workflow
• Batch Processing: Support for processing multiple texts
• Results History: View and manage previous processing results
• API Extension: Add support for more AI providers and models