Tim Berners-Lee Blog

We describe the development of “Charlie,” a AI assistant working for the user by leveraging personal data securely stored in a Solid Pod. Building on a 2017 proposal for a trustworthy and user-centric AI, the work highlights progress made by engineers at Inrupt in late 2024. Using a simulated dataset for a fictional user, “Bob,” the team integrated an advanced large language model (LLM) with Bob’s personal data to demonstrate the transformative potential of combining AI and SoLiD pods.

By accessing rich, structured, and personalized data, Charlie provided responses far superior to generic AI systems, as shown in a use case involving running shoe recommendations tailored to Bob’s fitness and lifestyle data. This personalization exemplifies the next level of AI capability, offering unprecedented usefulness while maintaining user trust.

We underscore the critical role of robust data infrastructure, including Solid Pods and the semantic web, in driving AI systems, and the dual role of AI in both populating and leveraging linked data stores, paving the way for a dynamic ecosystem where data graphs mediate AI interactions. We paint an exciting vision for integrating personal data, semantic web principles, and advanced AI to create tools that truly serve users.