TealKit

TealKit: A Privacy-First, Infinitely Extensible AI Platform for Mobile & Desktop

TealKit transforms your devices into a robust agentic AI platform, featuring autonomous agents, built-in tools, and limitless extensibility. Users can easily create personalized tools utilizing JavaScript, Python, PowerShell, or Bash, or connect to any MCP (Multi-Channel Platform) server, empowering AI with autonomous capabilities. TealKit is provider-independent, fully customizable, and prioritizes user privacy.

Recent Updates

• Server Mode: Operate the TealKit Server as a headless daemon on any always-on Linux device, including NVIDIA Jetson Nano, Mac Mini M4 Pro, Raspberry Pi, or cloud VMs. Connect your mobile or desktop application securely via API keys while allowing tasks and automations to function seamlessly in the background.
• Embedded Models: Download and run GGUF models directly on your device. This feature supports full offline functionality within the TealKit Server, allowing operations like text formatting, translation, and summarization without relying on external API keys or internet connectivity.
• Enhanced LLM Parameters: Now configurable within the global settings, this allows better control over aspects of model performance, critical for tasks requiring output consistency and quality.
• Chat Mode: This new feature allows users to bypass system prompts and tools, directly sending messages to the LLM, which optimizes performance for text-centric tasks.
• Agent Chaining: Create conditional and unconditional agent chains for refined task execution, allowing for added levels of control and complexity in workflows.
• Stop After Tool Call: This option allows agents to complete a single tool call before halting, facilitating immediate integration of tool output into subsequent processing without additional LLM formatting.

Upcoming Features

• Multi-server Control: Future releases will enable control of multiple headless server instances from a single mobile interface.
• Local Report Generation: Users will be able to generate charts and reports on headless server environments without requiring a graphical UI.
• Custom Trainers for Embedded Models: Tools for creating personalized domain-specific embedded models will soon be accessible, aiding customized functionality.

Core Capabilities

• AI Playground & Flexibility: Integrates with major providers such as Google Gemini, OpenAI GPT-5, and Mistral, ensuring compatibility and adaptability across diverse use cases.
• Data Privacy Compliance: All user data processing for European clients can be done through Mistral AI, maintaining full GDPR compliance.
• Small Language Models (SLM): Run lightweight SLMs locally for tasks that do not require extensive computational resources, preserving data privacy and minimizing cloud costs.
• Dual-LLM Support: Utilize a second LLM for specific tasks, allowing greater versatility in model selection and application.
• Local Intelligence: Enables 100% offline processing support for Ollama models and more.

Built-in AI Skills

• Document Intelligence: Local indexing of various document formats for efficient retrieval.
• Digital Office Functionality: Deep integration with tools like Gmail, Google Calendar, and universal IMAP/SMTP support.
• Content Creation & Visualization: Generate various file types and create diagrams from structured data.
• Remote Management: Execute server management tasks through SSH across platforms.

Extensibility Features

TealKit is an open agentic platform where users can create custom AI skills and connect to third-party MPC servers effortlessly. Supported programming languages include JavaScript, Bash, PowerShell, and Python, allowing a broad range of automated tasks to be scripted directly within the application.

Platform Availability

TealKit is supported across multiple platforms:

• Android: Access via Google Play Store.
• Windows: Available through Microsoft Store.
• Coming Soon: iOS, Linux, and Mac support.

Privacy Focused

TealKit ensures that user data remains secure and private, operating without the need for cloud data storage, hence maintaining total control over personal information and computational practices.