Many companies are switching to automated solutions as technology develops quickly and away from human customer care methods. Demand for conversational agents such as chatbots has so skyrocketed. Using natural language processing methods, these bots replicate human-like interactions with users via text or speech. In this respect, chatbots face two major challenges: intent classification and entity extraction from user requests. Therefore, this paper focuses on classification of intents along with named entity-identifying techniques to understand user requests. Understanding these key elements helps us improve the functionality and effectiveness of these conversational agents in meeting customer needs. The question-and-answer Amazon office products dataset is used in this paper. In order to achieve this objective, we employ the BiLSTM model in conjunction with deep learning to implement the intent classifier. Through bidirectional data processing, the model may better comprehend the relationship between sequences. In order to achieve a more precise and contextualized NER extraction, we implement Custom NER, which is based on BiLSTM. The custom NER achieved 96.94% accuracy, while the BiLSTM intent classifier model obtained 94.75% accuracy, according to the results.