In this Model HQ demo, I’ll show you how to build a no-code agentic workflow for sentiment analysis, emotion detection, ratings, and topic classification—all running locally on your device for fast, private, on-device AI. We start with an image input, use a vision model to describe what’s in the image, then stack sentiment + emotion models to classify the tone and feeling. Next, we pass the results through Agent State and add a chat model to generate a short story based on the vision + sentiment + emotion output. Then we switch to a text-based workflow and use sentiment + emotions to automatically generate a polite response to an angry customer message—a real-world use case for support teams, operations, and anyone handling inbound feedback. 🚀 Everything runs offline once models are downloaded — no Wi-Fi required. 🔒 Private, local, and easy to build with Model HQ’s visual agent builder. Subscribe for more how-to videos and visit llmware.ai for more information!