How to Drive AI Efficiency in a Consumption-Based World

As legal AI adoption expands, the cost of using AI is becoming a commercial question as much as an engineering one. In consumption-based models, the volume and complexity of AI use can have a direct impact on cost. Legal work, with its dense documents, complex reasoning and need for reliable outputs, presents particular challenges. AI effectiveness depends on three factors: the model, the harness and the context. While much of the focus has been on models and applications, the information supplied to AI is equally important. Context engineering is the process of making relevant legal knowledge available to AI in a form it can use. This session explores how a legal context graph can connect documents with the relationships, matter history and institutional knowledge that give them meaning. It examines why better context can help AI work more efficiently, and where the next opportunities to improve the economics of legal AI may lie.