Here is a neutral, objective, and interesting summary of the text: The latest release of Apache Flink 1.20 has brought several significant improvements to its usability, performance, and feature set, setting the stage for the upcoming major release of Flink 2.0. The update introduces a unified way to specify bucketing strategies for tables, making it easier to manage data distribution and freshness requirements. Additionally, Flink now supports computing aggregations on non-keyed streams in parallel and offers improved state management with multiple configurations to control upload behavior. The community has also fixed several issues, including a longstanding problem with RocksDB state backend, and made final preparations for the upcoming Flink 2.0 release. Furthermore, Confluent Cloud for Apache Flink supports AI model inference and enables the use of models as resources in Flink SQL, expanding its capabilities for machine learning and data processing.