Assessment and feedback often come up (for example in NSS surveys) as an area that can be improved on within mathematics and statistics degrees. Assessment in the mathematical sciences is predominantly by time-restricted unseen exams. Over the past decade there has been considerable effort to use a more varied and innovative assessment strategies, including open-book exams and project- or portfolio-based coursework. These new assessment methods have been motivated variously by the necessity of running remote assessments during a pandemic, to broadening participation in the mathematical sciences, or to provide more authentic assessment in addition to students' ability to problem-solve under time pressure. However, the advent of AI means that assessment of mathematics via coursework is becoming ever more challenging. Exams are a cause of a lot of stress and anxiety for students, and the workload associated with assessment and feedback can feel unmanageable for both staff and students. In this workshop we will explore how we could do this better within mathematics, looking at innovative approaches to assessment and feedback. Talks will cover a range of topics, including using AI for feedback on proofs, formative feedback in large lecture classes, and unintended consequences of traditional assessments in mathematics.