ARCHIVES

Modern web applications are highly dynamic, making traditional automation tools fragile due to their reliance on static scripts and DOM selectors. This paper proposes an AI-powered browser navigation agent that integrates Large Language Models (LLMs) with the Skyvern framework to enable adaptive and intelligent automation. The system interprets natural language instructions, generates execution plans, and interacts with web pages using visual understanding rather than brittle selectors. Experimental results demonstrate improved accuracy, flexibility, and reduced manual effort compared to conventional automation systems