Lance is a modern columnar data format implemented in Rust, designed to provide fast random access performance, especially beneficial for machine learning applications such as search engines, real-time feature retrieval, and deep learning training. In benchmarks comparing Lance to Parquet, Lance demonstrated significantly faster random access, achieving performance nearly 2000 times quicker. Unlike other systems that require complex setups or costly in-memory stores, Lance is easy to use with a simple installation process. The format is especially suited for scenarios requiring efficient data access and manipulation, and it is actively being developed to include features like generalized key lookups and improved DuckDB integration. Lance is open-source and invites feedback and contributions from the community to enhance its capabilities further.