Performance, Scalability and Maintenance: An Experimental Study of RPA vs. Autonomous Agents in the Healthcare Domain

Comparing RPA using Playwright against LLM-based autonomous agents (LangGraph and Gemini 2.5 Flash-lite) for medical data retrieval shows RPA maintains higher accuracy (98.90% vs 96.49% F1) and significantly lower costs. RPA performed 13 times faster, averaging 16.07 seconds per patient compared to 207.21 seconds for the APA agent.