AWS describes how Amazon Quick and Amazon Bedrock Knowledge Bases enforce document-level permissions in enterprise RAG over sources like SharePoint, Google Drive, and Confluence. The system uses a hybrid two-stage design: pre-retrieval ACL filtering on a cached vector index, followed by real-time permission verification with the authoritative source at query time. This addresses stale or incorrectly replicated ACLs that could leak sensitive documents through AI answers.
A cybersecurity student released GateKeep RAG, an open-source framework treating multi-tenant RAG isolation as a database and retrieval problem instead of a prompt-guardrail problem. It enforces authorization at the vector-search and relational layers, with an audit ledger and adversarial evaluation showing zero leaks across 823,996 canary checks. The work highlights that prompt instructions fail as authorization boundaries once unauthorized chunks enter an LLM's context.
A cybersecurity student released GateKeep RAG, an open-source framework treating multi-tenant RAG isolation as a database and retrieval problem instead of a prompt-guardrail problem. It enforces authorization at the vector-search and relational layers, with an audit ledger and adversarial evaluation showing zero leaks across 823,996 canary checks. The work highlights that prompt instructions fail as authorization boundaries once unauthorized chunks enter an LLM's context.
A YouTube talk features Aaron Linsky (Bridgewater) and Nikhil Benesch (turbopuffer) explaining how Bridgewater's Vault — 100 million documents with per-document permissions — broke conventional filtered vector search, and how turbopuffer's 'native filtering' fixed it. The discussion is a practitioner case study in making retrieval-augmented AI respect document-level access control at scale.
A YouTube talk features Aaron Linsky (Bridgewater) and Nikhil Benesch (turbopuffer) explaining how Bridgewater's Vault — 100 million documents with per-document permissions — broke conventional filtered vector search, and how turbopuffer's 'native filtering' fixed it. The discussion is a practitioner case study in making retrieval-augmented AI respect document-level access control at scale.
AWS published two posts on building agentic retrieval with Managed Amazon Bedrock Knowledge Bases: one on observability for enterprise agentic retrieval using AWS CloudFormation, and one on multi-tenant agentic chat over enterprise data. Both target the step beyond single-source RAG, where an agent routes questions across knowledge bases and retrieves iteratively. The multi-tenant post emphasizes per-tenant document isolation enforced from verified identity rather than client-supplied values.
AWS published two posts on building agentic retrieval with Managed Amazon Bedrock Knowledge Bases: one on observability for enterprise agentic retrieval using AWS CloudFormation, and one on multi-tenant agentic chat over enterprise data. Both target the step beyond single-source RAG, where an agent routes questions across knowledge bases and retrieves iteratively. The multi-tenant post emphasizes per-tenant document isolation enforced from verified identity rather than client-supplied values.
Amazon announced the general availability of the Amazon Quick desktop application for macOS and Windows, an enterprise AI assistant that runs on AWS with data staying in the customer's environment. It also added a mobile activity feed for iOS and Android that consolidates email, calendar, CRM, and messaging into one prioritized view while background agents resolve routine items. Customers cited include Southwest Airlines, LabCorp, and PGA TOUR.
Amazon announced the general availability of the Amazon Quick desktop application for macOS and Windows, an enterprise AI assistant that runs on AWS with data staying in the customer's environment. It also added a mobile activity feed for iOS and Android that consolidates email, calendar, CRM, and messaging into one prioritized view while background agents resolve routine items. Customers cited include Southwest Airlines, LabCorp, and PGA TOUR.
An arXiv paper presents STAIR (STructure Aware Information Retriever), a dataset and LLM-based retriever that augments RAG with document structure information. It targets the 'lost in the middle' problem by preserving the semantic global structure that length-based chunking discards.
An arXiv paper presents STAIR (STructure Aware Information Retriever), a dataset and LLM-based retriever that augments RAG with document structure information. It targets the 'lost in the middle' problem by preserving the semantic global structure that length-based chunking discards.