cookbook/README.md
This repository contains comprehensive demos and examples for Nexa SDK across different platforms: PC (Python), Android, and Linux.
All PC demos are located in the PC/ directory and run on Windows (x64/ARM64 Snapdragon), macOS, and Linux.
Lightweight on-device AI assistant with function calling (web search) using Granite-4-Micro model. Includes Gradio interface for local agentic workflows.
Function calling capabilities with NexaAI VLM model, integrated with Google Calendar via MCP protocol. Supports multi-modal input (text, image, audio) with Web UI and CLI interfaces.
End-to-end Retrieval-Augmented Generation pipeline with embeddings, reranking, and generation models. Query your own documents (PDFs, Word, text) locally on device.
Vision-Language Model (VLM) with structured JSON outputs, multi-image input, and custom system prompt control.
Lightweight RAG system with Qwen3VL multimodal model powered by Nexa Serve. Supports PDFs, Word docs, text files, and images. Includes CLI and Gradio UI.
Local multimodal LLM-driven browser automation using Qwen3-VL. Enables autonomous web navigation, searches, and complex web tasks.
For Android SDK demos and examples, please refer to android/.
Linux-specific demos coming soon in the linux/ directory.