Calculated columns are now available in preview for Direct Lake on OneLake semantic models, addressing a key user request. This feature allows the creation of derived columns directly within the semantic model using DAX. Previously, this required moving data to a different storage mode or pushing logic upstream. Direct Lake offers Import-like performance with near real-time data freshness by querying Delta tables. Common use cases include deriving values, concatenating fields, formatting dates, and personalizing content. Calculated columns in Direct Lake support an "Expression Context" property, evaluated at query time and respecting user security. This means they automatically adhere to row-level and object-level security. Unlike standard calculated columns, Direct Lake versions are evaluated within the user's security context. This prevents sensitive data from being exposed if the user lacks permissions. However, Direct Lake calculated columns cannot be used in relationships as they do not materialize. They also do not observe filter context, meaning their results are fixed per row. Measures should be used when dynamic results based on report interactions are needed. To start, open a Direct Lake on OneLake semantic model and create a calculated column.