Developed and delivered a Web Agent Demo for the NexaAI/nexa-sdk repository, showcasing integration of the Qwen3-VL multimodal model to enable local execution of complex web tasks. The project combined Bash scripting and Markdown documentation to provide clear setup instructions, usage examples, and feature highlights, supporting rapid prototyping and demonstration scenarios. By connecting the Nexa SDK with a user-facing web UI, the work demonstrated autonomous browser automation powered by vision-language understanding. The focus on demo setup, web automation, and comprehensive documentation facilitated onboarding for new users and highlighted the end-to-end capabilities of the underlying technologies within a single feature release.