Abstract: The Parliamentary Intelligence Agent (PIA) is a real-time NLP dashboard designed to support investigative journalism by monitoring South African parliamentary activity at scale, built for Athandiwe Saba of the Daily Maverick. PIA addresses a practical newsroom problem: the Parliamentary Monitoring Group publishes hundreds of committee transcripts, bills, and parliamentary questions each month. PIA fetches live data directly from the PMG REST API across three streams, 576 committee meeting transcripts, 21 active bills, and 2,175 questions and replies, and applies a multistage NLP pipeline comprising BART abstractive summarisation, spaCy named entity recognition, DistilBERT sentiment scoring, and LDA topic modelling. An eight-category rule-based alert engine generates 166 prioritised signals surfaced through a 12-tab Streamlit dashboard. A Story Lead Generator tab uses DeepSeek-V3 to produce AP-style investigative leads directly from the top alerts. Validation against the PCS framework yields 7/7 checks passed with a mean score of 0.92. The system is deployed on Streamlit Community Cloud via a public GitHub repository with PMG permission.