Abstract Proposal: This session shares observations from an experiment using agentic AI in a mobile forensic case involving a timestamp interpretation issue. It examined whether an AI agent would simply trust parsed forensic tool results or perform a more comprehensive review. Rather than making broad claims about AI capability, the session focuses on what happened when an AI agent was used for context-building, artifact review, and validation. Attendees will learn about potential risks in AI-assisted analysis, ways to use agents more effectively, and why human analyst oversight remains critical.