The availability of powerful open-source large language models (LLMs) opens exciting use-cases, such as automated personal assistants that adapt to the user's unique data and demands. Two key desiderata for such assistants are personalization-in the sense that the assistant should reflect the user's own style-and privacy-in the sense that users may prefer to always store their personal data locally, on their own computing device. We present a new design for such an automated assistant, for the specific use case of personal assistant for email generation, which we call Panza. Specifically, Panza can be both trained and inferenced locally on commodity hardware, and is personalized to the user's writing style. Panza's personalization features are based on a new technique called data playback, which allows us to fine-tune an LLM to better reflect a user's writing style using limited data. We show that, by combining efficient fine-tuning and inference methods, Panza can be executed entirely locally using limited resources-specifically, it can be executed within the same resources as a free Google Colab instance. Finally, our key methodological contribution is a careful study of evaluation metrics, and of how different choices of system components (e.g. the use of Retrieval-Augmented Generation or different fine-tuning approaches) impact the system's performance.

Data and analytics capabilities are becoming increasingly important'table stakes' in the property and casualty sector across Europe, North America and Asia, according to a session at the Barbados Risk and Insurance Management (BRIM) conference. Speaking at the session'Next generation insurtech: predictive modelling and artificial intelligence (AI)', Klaas Stijnen, co-founder and chief product officer at Montoux, cited NewVantage Partners' 2022 Big Data and AI Executive Survey, which found that although investment in data and AI initiatives continues to grow, achieving data-driven leadership remains an elusive goal for most organisations. Similarly, the survey found that although the take-up of AI initiatives is accelerating, the actual implementation of AI into widespread production remains low. Stijnen outlined that a data analysis-driven approach is led by data scientists, in which the focus is on available, known data and is separate from the decision-making process. Alternatively, a decision-driven approach is based on data science, which more readily challenges bias to seek missing data and is integrated into a firm's decision-making process.