6 posts found
MCP and function calling both let an AI use tools. But function calling is a per-request contract you define in code, while MCP is a persistent protocol that exposes tools from a running server. This guide explains the difference clearly, when each approach makes sense, and how tools like Askimo Desktop let you use both without writing any code.
MCP (Model Context Protocol) lets your AI do real things, not just answer questions. Connect Askimo Desktop to GitHub, databases, local files, web search, and any API via Stdio or HTTP. Secrets stay in your OS keychain. Works with Claude, GPT, Gemini, and local Ollama models.
New in Askimo: built-in terminal emulator, Model Context Protocol (MCP) integration, a RAG sources tree to inspect exactly what the assistant used, image-model support for multimodal workflows, and major desktop improvements across performance, reliability, and UX.
A comprehensive technical deep dive into building Askimo - an open-source desktop AI chat client for ChatGPT, Claude, Gemini, and Ollama built with Kotlin and Compose. Learn how we architected support for multiple AI models, handled concurrent streaming, implemented RAG for documents, and solved real-world challenges in production.
The best AI desktop app isn't the one with the best single model - it's the one that lets you use ChatGPT, Claude, Gemini, and Ollama together. Switch models mid-conversation, keep sensitive data local, and cut API costs by routing simple tasks to free local models. Askimo is free, open source, and works on macOS, Windows, and Linux.
Askimo App complements the CLI for interactive AI work. Switch between OpenAI, Claude, Gemini, and Ollama in one session without separate browser tabs. The app maintains your workspace and conversation context when switching providers. Start with OpenAI, jump to Gemini, or use local Ollama models seamlessly. Features offline support, secure storage, and native cross-platform performance for unified AI workflows.
