AWS published a reference architecture that pairs Amazon Quick with fixed operations and a rules engine to check leases and document coverage.
AWS published a technical guide on October 2, 2026, describing an “Adjudicated Query” pattern for reviewing large lease portfolios through Amazon Quick. The approach is designed for compliance tasks where teams need both an exhaustive check and evidence showing what was assessed. AWS frames a portfolio of 50,000 leases across multiple states as the scale of the challenge, with changing landlord-tenant rules such as late-fee caps and notice periods.
The chat interface translates a user’s question into one of a fixed set of operations and explains the returned result. A deterministic rules engine—not the language model—makes pass-or-fail compliance decisions. AWS says the model cannot write queries, choose which records count toward the population, or make the determination. The system also distinguishes exhaustive checks from exploratory clause searches and proposed rule-change simulations.
In the reference architecture, Amazon Quick communicates with a bounded Model Context Protocol server through Amazon API Gateway and AWS Lambda. Amazon Cognito handles authentication, while Amazon Aurora Serverless v2 stores lease records, rules, findings and audit information. Amazon Quick Sight provides a dashboard for inspecting the full result set. Amazon Bedrock is used only for exploratory semantic search and qualitative clause assessment, not for official compliance sweeps.
AWS demonstrates the design with synthetic data. In its Texas late-fee example, the reported sweep found 10,111 violations, 689 ambiguous cases and 20 unreadable records. A completeness receipt accounts for records across outcome categories, while the dashboard provides lease-level findings, including the rule applied, compared values and supporting clause text. The sample includes 28 acceptance checks and can be deployed with an AWS Cloud Development Kit stack.
The guide emphasizes that the pattern is suited to enumerable populations and rules that can be expressed as deterministic comparisons. It cautions that the method does not resolve subjective legal judgments, and that a completeness receipt proves coverage of the selected population—not that the population itself was defined correctly. The sample uses synthetic records and placeholder rules, so AWS says organizations must validate legal content and data practices before applying it to real leases.