This paper presents an integrated AI-based pipeline for law enforcement applications, specifically targeting the creation of first information report (FIR) in the Kannada language. The pipeline comprises a speech processing block utilizing automatic speech recognition (ASR) models to convert spoken Kannada into text, and a natural language processing block employing advanced AI techniques for information extraction and generating concise FIR summaries. The research focuses on optimizing and streamlining the FIR creation process by leveraging AI technologies and overcoming data scarcity in the Kannada language through data augmentation techniques. OpenAI’s Whisper is used for ASR, resulting in two-word error rate in this use case. For information extraction, a Bi-LSTM neural network is employed for named entity recognition (NER) with a performance of 94%. The proposed NER model utilizes three tags, which can be expanded to accommodate additional tags as per FIR requirements.