TealKit
TealKit is a privacy-first AI agent app for mobile and desktop. Run manual or scheduled AI workflows, connect MCP tools, and use cloud or local models.
TealKit is an agentic AI platform for mobile and desktop focused on privacy, flexibility, and automation.
Core capabilities
AI Playground and Flexibility
Provider independent: works with Google Gemini, OpenAI GPT-5, Anthropic, Mistral, and OpenAI-compatible endpoints.
EU privacy support: works with Mistral AI for EU data residency and GDPR-focused deployments.
Local and SLM support: use Ollama, LM Studio, and embedded GGUF models for low-cost or offline usage.
Dual LLM setup: configure LLM 1 and LLM 2 and switch quickly in the Playground.
Chat mode and task mode: test prompts in chat, then promote to reusable tasks.
Advanced controls: tune model settings per task for cost, latency, and quality.
Auto system prompt: generate and edit a tool-aware system prompt before execution.
Built-in AI Skills
Document intelligence: local RAG with DuckDB for PDF, Word, and Excel semantic search.
Digital office: integrations for Gmail, Google Calendar, Google Drive, and IMAP/SMTP workflows.
Content and visualization: generate TXT, PDF, Excel, and Mermaid-based flowcharts.
Web search and indexing: search with SerpApi or DuckDuckGo and crawl websites for local indexing.
Remote management: SSH across platforms plus desktop Bash and PowerShell script execution.
Agentic Workflows and Automation
Autonomous scheduling: cron-based agents that run in background.
Agent chaining: build multi-step pipelines with optional conditions.
Stop-after-tool flow: pass raw tool output to next agent without extra LLM rewrite.
Multi-channel output: save locally or send via email, Slack, or WhatsApp with attachments.
Tip for submission
Use only plain ASCII text, no emojis, no markdown tables, and keep the long description under about 2500 to 3500 characters unless the form states a higher limit.
Core capabilities
AI Playground and Flexibility
Provider independent: works with Google Gemini, OpenAI GPT-5, Anthropic, Mistral, and OpenAI-compatible endpoints.
EU privacy support: works with Mistral AI for EU data residency and GDPR-focused deployments.
Local and SLM support: use Ollama, LM Studio, and embedded GGUF models for low-cost or offline usage.
Dual LLM setup: configure LLM 1 and LLM 2 and switch quickly in the Playground.
Chat mode and task mode: test prompts in chat, then promote to reusable tasks.
Advanced controls: tune model settings per task for cost, latency, and quality.
Auto system prompt: generate and edit a tool-aware system prompt before execution.
Built-in AI Skills
Document intelligence: local RAG with DuckDB for PDF, Word, and Excel semantic search.
Digital office: integrations for Gmail, Google Calendar, Google Drive, and IMAP/SMTP workflows.
Content and visualization: generate TXT, PDF, Excel, and Mermaid-based flowcharts.
Web search and indexing: search with SerpApi or DuckDuckGo and crawl websites for local indexing.
Remote management: SSH across platforms plus desktop Bash and PowerShell script execution.
Agentic Workflows and Automation
Autonomous scheduling: cron-based agents that run in background.
Agent chaining: build multi-step pipelines with optional conditions.
Stop-after-tool flow: pass raw tool output to next agent without extra LLM rewrite.
Multi-channel output: save locally or send via email, Slack, or WhatsApp with attachments.
Tip for submission
Use only plain ASCII text, no emojis, no markdown tables, and keep the long description under about 2500 to 3500 characters unless the form states a higher limit.
