The rapid growth in the healthcare industry demands innovative techniques to extract a sufficient number of required information from unstructured textual information, for example, clinical notes, patient records, research papers, and diagnostic reports. Such bottlenecks delay the time of decision-making and proper care delivery. Therefore, we have developed a Hybrid Generative AI Framework combining both extractive and abstractive summarization approaches using rule-based, supervised, and unsupervised learning methods. This framework includes all the leading AI models like BioBERT, architectures of GPT, and domain ontologies namely UMLS and SNOMED-CT for the production of summaries that are not only trustworthy and contextually relevant but factually accurate with the inclusion of NER, TF-IDF, and optimization based on reinforcement learning, thus beating the contextual limitation of traditional methods and factual inaccuracy of abstractive methods. The performance was measured in depth using metrics like ROUGE-1, ROUGE-2, ROUGE-L, BLEU, execution time, and memory usage. Results were found to be highly improved, such as 93.2% NER precision and a 92.6% reward optimization score, which assures the suitability of the framework in generating clinically relevant summaries. The hybrid approach also showed efficiency in terms of execution time and balanced memory usage, thus being scalable for real-time applications. It allows for an easier and much quicker appraisal of the process by medical professionals, subsequently making for better clinical decisions to result in better health outcomes of patients. As such, it becomes the benchmark by which medical literature is summarized to provide guiding principles toward improving healthcare provision and improving precision medicine.