To enhance person-centred care, information about patients clinical needs, their service encounters, and their own preferences should be shared across multidisciplinary teams (MDTs). Too often, this information is neither adequately recorded nor shared. Despite the limited time patients spend with their GPs(1) and the historic amounts of pressure the National Health Services (NHS) and GP practices are facing(2), important conversations are happening with GPs every day, which alone have been reported to be healing(3). Artificial intelligence (AI) models that automatically summarise these encounters can help alleviate his pressure. Unfortunately, existing models were developed to mimic current summarization practices of doctors that focus more on clinical needs or issues and less on patients preferences, values, and concerns(4). The aim of this project is to test if our AI model can overcome this limitation and develop person-centred post-encounter summaries. We have developed a more explainable AI-model that automatically generates summaries from multiple documents or dialogues(5). Our model was developed by using large language models (i.e. LongT5) and it was validated on 165,000 set of webpages and 3,673 transcripts of dialogues from TV episodes. The difference between this model and other summarization models (e.g. GPT-4) is that our model has an intermediate planning step that allows us to explain and personalize the automatic summarization process. However, our model has not yet been validated on clinical settings. Our proposal, which aligns with the FAST priority of AI for primary care settings, is to test and validate our AI summarization model on the generation of person-centred summaries of patients-GPs clinical encounters in real-life scenarios. To test if these summaries are person-centred clinical summaries, we will use Patient and Public Involvement (PPI) groups throughout our study.