Skip to main content

Showcase Applications

This section highlights real-world applications built using Airtrain. These examples demonstrate how developers are using Airtrain to create powerful AI-powered solutions across various domains.

TRMX​

TRMX is a powerful terminal-based chat interface that enables users to interact with various AI models and build personal AI agent and run locally directly from their command line. It provides a seamless way to work with multiple AI providers while maintaining privacy through local storage of conversations.

Key Features​

  • Multi-Provider Support: Seamlessly integrates with OpenAI, Anthropic, TogetherAI, Groq, Fireworks, Cerebras, and Google AI
  • Dynamic Model Selection: Lists and selects from all available models across providers with automatic API-based discovery
  • Local Storage: All conversations are stored locally on the user's machine for complete privacy
  • Session Management: Maintains context and history across conversations with intelligent session titling
  • Advanced Multi-Line Input: Supports complex prompts with multiple input methods including paste mode
  • Markdown Rendering: Rich formatting of AI responses in the terminal with syntax highlighting
  • Thinking Process Visualization: Option to view the model's thinking process for supported models
  • Intuitive CLI Interface: User-friendly command-line interface with clear organization of outputs
  • Automatic Updates: Built-in version checking and self-update functionality
  • Time Display Options: Flexible timestamp formatting with absolute, human-readable, and relative options
  • Token Estimation: Built-in token counting to help manage context limits
  • Intelligent Title Generation: Automatic creation of relevant session titles based on conversation content
  • Credential Management: Secure storage and handling of API keys for multiple providers
  • Internet Connectivity Checks: Graceful handling of connectivity issues during API calls

How TRMX Uses Airtrain​

TRMX leverages Airtrain's agent framework to create specialized agents that understand terminal commands, filesystem operations, and command-line tools. The agents use:

  • Memory Systems: To retain context across terminal sessions and manage both short-term working memory (shared memory between agents) and long-term persistent memory (Indexed persistent memory)
  • Schema Validation: To ensure safe execution of suggested commands with quick validation of inputs and outputs
  • Permission Systems: To control what actions the AI can take on the system
  • Multi-Model Integration: To seamlessly switch between different AI models based on the task requirements
  • Local LLM Support: To provide offline capabilities through integration with Ollama and other local models
  • RAG Integration: To efficiently retrieve relevant information using LanceDB for contextual understanding
  • Provider Integration Framework: Uses Airtrain's standardized integration system to seamlessly work with multiple AI providers
  • Model Discovery: Leverages Airtrain's ListModelsSkillFactory to dynamically discover available models for each provider
  • Credentials Management: Implements Airtrain's credential storage and retrieval patterns for secure API key handling
  • Standardized Message Formats: Uses Airtrain's consistent message formatting system across different providers
  • Dynamic Module Loading: Employs Airtrain's module system to load provider-specific implementations at runtime
  • Error Handling: Integrates Airtrain's robust error detection and recovery mechanisms
  • Provider-Specific Input Classes: Utilizes Airtrain's input validation and formatting for different provider requirements
  • Auto-Adaptation: Dynamically adapts to provider-specific requirements and limitations through Airtrain's abstraction layer
  • Consistent Output Processing: Standardizes the varied responses from different AI providers into a uniform format

Zchat​

Zchat is a lightweight, privacy-focused CLI chat agent that allows users to have conversations with AI models completely from their terminal. It stores all conversation data locally on the user's machine, ensuring privacy and data control.

Key Features​

  • Command Line Interface: Simple 'z' command for interacting with AI assistants directly from the terminal
  • Local Storage: All conversations are stored locally on the user's machine for complete privacy
  • Session Management: Start new chats or continue existing conversations with intuitive commands
  • Multi-Provider Support: Compatible with multiple LLM providers through Airtrain's integrations
  • Multi-Line Input: Support for complex, multi-line prompts with several input methods
  • Conversation History: Easy access to past conversations with search and continuation capabilities
  • Offline Use: Can be configured to work with local LLMs for offline use
  • Markdown Rendering: Beautiful rendering of AI responses with rich formatting in the terminal
  • Credential Management: Secure storage and handling of API credentials
  • Internet Connectivity Checks: Graceful handling of connectivity issues
  • User-Friendly Interface: Rich text formatting with colored output for better readability
  • Simple Installation: Available through PyPI with minimal dependencies

How Zchat Uses Airtrain​

Zchat leverages Airtrain's capabilities to provide a seamless chat experience:

  • Provider Integration: Utilizes Airtrain's pre-built integrations with LLM providers (e.g., TogetherAI)
  • Message Formatting: Uses Airtrain's standardized message formats for consistent provider communication
  • Session Persistence: Implements a file-based storage system that saves conversation history in JSON format
  • Credential Management: Leverages Airtrain's credential handling mechanisms for secure API key storage
  • Configuration System: Built on Airtrain's configuration patterns for flexible settings management
  • Error Handling: Implements robust error detection and recovery using Airtrain's error patterns
  • Model Selection: Supports dynamic model selection through environment variables
  • Rich Output: Integrates with terminal UI libraries to provide beautifully formatted responses

Building Your Own Showcase App​

These applications demonstrate how Airtrain's flexible architecture enables developers to build sophisticated AI-powered applications. To showcase your own application built with Airtrain, please submit a pull request to our documentation repository with details about your application and how it uses Airtrain's capabilities.

For guidance on building production-ready applications with Airtrain, refer to our Advanced Examples and API Reference sections.

If you're using Airtrain in your project, we'd love to feature it here! Fork our repository and raise a PR to add your application to the showcase.