Patent portfolio reviews remain one of the most important and resource-intensive exercises for IP teams, yet many maintenance decisions are still driven by traditional indicators such as citations, family size, patent age, and expert judgment. This webinar explores a new approach: assessing portfolio value based on evidence of real-world use. Leveraging advances in large language and multimodal AI, organisations can now analyse patent claims at scale, identify products and technologies that appear to practise those claims, and generate maintenance scores grounded in market activity rather than proxy metrics. The session will examine how these systems work, the opportunities they create for portfolio optimisation, and the limitations practitioners need to understand, including detectability challenges, evidentiary gaps, and the risk of incorrect conclusions when assessing patent value.