We present PodChecker, a user-facing system for automated, claim-level fact-checking of podcast content. PodChecker processes podcast audio or RSS feeds by transcribing episodes, extracting atomic factual claims, and assigning each claim one of four fine-grained labels - true, false, misleading/partially true, or unverifiable - using retrieval-augmented verification. The system presents fact-checking results at the level of individual claims, accompanied by simple visual indicators and links to supporting/conflicting sources. This design, implemented via an interactive web-based interface, enables users to inspect fact-checking outputs and underlying evidence directly, supporting interpretable and critical engagement with long-form audio content. By presenting claim-level evidence and labels, PodChecker assists both general listeners and professional fact-checkers in assessing podcast factuality.

PDF