We’re constantly improving how Ragie handles retrievals, ensuring that you get the most relevant and up-to-date data when you need it most. Today, we’re excited to unveil two significant upgrades to Ragie’s retrieval capabilities that make accessing critical insights faster and more accurate.

What’s New?

• Recency Bias: In verticals like sales and customer support, where the most recent data is often more accurate, traditional retrieval systems can fall short by treating all documents equally, regardless of when they were created or updated. With Recency Bias, Ragie introduces time-aware weighting that promotes fresh content. This means your applications can stay aligned with the ever-changing data and surface only the most up-to-date information when it matters most. Learn more about how to implement Recency Bias in our API docs. 
• Specialized Table Chunking: Complex documents, such as financial reports or legal contracts, are often filled with tables, charts, and figures that contain valuable insights.  However, parsing and retrieving this information accurately can be challenging. With our new Specialized Table Chunking, Ragie now offers enhanced extraction capabilities, ensuring that table headers remain correctly associated with their rows for more precise and contextual data retrieval.  

To see precisely how Ragie stacks up in real-world applications, visit our new benchmarks page, where we dive into how these new features, alongside many others, have helped Ragie outperform rigorous tests like LegalBench-RAG and FinanceBench.