Amazon Bedrock adds tabular search for agentic retrieval
In short: Amazon Bedrock Managed Knowledge Base now supports tabular search, allowing agents to process data directly from CSV, XLSX, and XLS files. The system uses model-written code in a secure sandbox to perform calculations and lookups across entire files, combining the results with facts from unstructured documents to provide comprehensive answers.
This summary was generated automatically by AI from AWS Machine Learning's publication. It is our own text, not a copy of the original — facts, figures and quotes belong to the source, linked above and below.
What changed?
- 1Native support for CSV, XLSX, and XLS files in knowledge bases
- 2Agentic retrieval now performs code-based calculations and lookups across full files
- 3Secure, isolated sandbox environment for running model-written code
- 4Combined reasoning across table data and text-based documents
Why it matters
This feature solves the common RAG failure point where tables are split into fragmented passages, causing loss of context for calculations. It is ideal for scenarios requiring cross-referencing of policies and quantitative data without complex manual ETL processes.
What it means for AI agents and contact centers
For your voice AI agent pipelines, this reduces the need to pre-process data into vector chunks or flatten Excel sheets. When testing, prioritize measuring the latency added by the extra agent turn and ensure the sandbox environment correctly handles your specific domain-specific data structures.
🧪 Worth Testing
Testing is recommended to see how it simplifies your agent's ability to handle multi-step analytical questions that mix policy documents with raw financial or operational data.
Sources
- AWS Machine LearningOfficialPrimary sourceOriginal article →„Get answers from tables with tabular search in Amazon Bedrock Managed Knowledge Base“16 Nov 2026, 18:12
- Published by source
- 16 Nov 2026, 18:12
- Found by our system
- 16 Nov 2026, 18:35
- Summary generated
- 16 Nov 2026, 18:36
This article was written by AI from the original source. Facts, numbers and prices come from the source; missing values are marked “Not specified”. Legal notice, copyright and privacy