Axelera® Development Systems with Arduino

Steven Hunsche, Senior Customer Engineering & Success at Axelera® AI, demonstrates how the Metis AI Processing Unit brings large language model capabilities to the edge, with zero cloud connection required. Using a carrier board with the Axelera AI accelerator alongside the Arduino Portenta X8, he showcases a live Llama-3.2-3B-Instruct chatbot demo that responds in real time to user questions, generating a detailed, context-aware response entirely offline. This hands-on walkthrough highlights how Axelera's edge AI acceleration makes private, low-latency, on-device language model deployment a practical reality, no internet connection needed.