Skip to content
context Productivity

Amazon Bedrock Managed Knowledge Base can search CSV and Excel tables

clearly official source

AWS has added a tabular search feature to Amazon Bedrock Managed Knowledge Base that searches CSV, XLSX, and XLS tables and combines the results with facts from documents using agentic retrieval.

Amazon Bedrock Managed Knowledge Base now supports tabular search – a feature that enables retrieving answers from tables (CSV, XLSX, XLS) as part of agentic retrieval. According to AWS, the feature runs calculations or value lookups across all relevant rows of one or more tables or sheets and combines the result with facts from related documents, which makes it possible to answer analytical questions requiring multiple steps.

Technically, the feature runs through the same AgenticRetrieveStream request already used for documents. The code runs over the original file in an isolated sandbox with no network access, so according to AWS the answer is based on the entire table, not just the rows returned by classic search. According to the article, the traditional RAG approach loses information when splitting tables into fragments (passages), and moreover has trouble linking business terminology from the question to codes and numbers in tables – tabular search solves this problem by having search only locate the correct file, after which the code reads the entire table.

The feature works on tabular files regardless of which connector was used to upload them into the knowledge base (Amazon S3, Microsoft SharePoint, Atlassian Confluence, and others), and does not require a schema definition. According to AWS, however, it is intended for queries requiring calculations or lookups over specific tables, not as a replacement for BI tools for governed metrics, recurring dashboards, or joins at the scale of a data warehouse. Running the code also adds one additional agent step to the response compared to purely text-based answers.

What changed

Why it matters

Companies that combine tabular data (budgets, P&L statements, BI exports) with text documents (policies, memos) in Amazon Bedrock Managed Knowledge Base can now get answers that link both layers in a single query without needing manual integration or schema definition. This shortens the work for analysts who previously had to manually combine data from tables with context from documents.

Two audiences, two different impacts

What this means

01

For individuals

For developers and analysts working with Amazon Bedrock Managed Knowledge Base, there is now an option to query CSV/Excel files via AgenticRetrieveStream and get an answer backed by calculations over the entire table as well as citations from documents, without having to build their own integration with a BI tool.

What to do Try the tabular search feature on sample data from the GitHub repository sample-managed-knowledge-base-patterns.
More practical updates →
02

For a business

Companies using Amazon Bedrock Managed Knowledge Base can deploy analytical queries combining tabular data and documents (e.g., financial closings, compliance checks) without building their own integration between RAG and tabular sources, which reduces the cost of implementing such internal tools.

Development
What to decide Consider whether deploying tabular search in Amazon Bedrock Managed Knowledge Base could simplify internal reports combining tables and documents.
More business impacts →
agentic retrieval Amazon Bedrock AWS knowledge base Retrieval-augmented generation

Check the original

Event sources

clearly official source · 1 publisher, 0 independent. We count feeds from the same owner only once.

1
AWS Machine Learning Blog primary source · first detected Get answers from tables with tabular search in Amazon Bedrock Managed Knowledge Base