What is Legal, Regulatory, Trade & Compliance Agent


The Legal, Regulatory, Trade & Compliance (LRTC) Analysis Agent is an intelligent compliance evaluation system designed to streamline and standardize the assessment of trade compliance requests. The agent automates the analysis of validated submissions by applying approved classification frameworks, regulatory rules, and duty determination logic to structured product and trade data. By leveraging authoritative regulatory sources exclusively, the agent ensures that all compliance evaluations, classifications, and advisory recommendations are accurate, consistent, traceable, and aligned with organizational and regulatory standards.

The agent supports the complete trade compliance workflow from intake validation through advisory output generation and final determination record creation. It evaluates product attributes, regulatory requirements, trade restrictions, and applicable duty considerations to produce structured determination outputs that assist reviewers and decision-makers during authorization and compliance review processes. Through controlled processing, auditability, and standardized logic execution, the LRTC Analysis Agent helps organizations reduce manual effort, improve decision consistency, accelerate compliance assessments, and maintain confidence in regulatory adherence across trade operations.

What the Agent Does


The Legal, Regulatory, Trade & Compliance (LRTC) Analysis Agent automates the intake validation and initial assessment of trade classification requests submitted through the approved intake interface. The agent receives validated request files along with supporting documentation and performs a structured verification process to ensure all required information is present before analysis begins. This includes validating product identifiers, product descriptions, country-related information, category-specific attributes, Decision Matrix responses, and all mandatory supporting documents in accordance with the Standardized Input Schema and Product Attributes Schema.

As part of its validation workflow, the agent verifies that supporting documentation is correctly associated with the relevant product or part number referenced in the request and confirms that the contents of each document align with its declared document type. The agent is designed to identify incomplete, unsupported, inconsistent, or invalid submissions early in the process. Where minimum intake validation requirements are not met, the agent automatically halts further processing to prevent inaccurate classifications or regulatory evaluations from proceeding through the compliance workflow.

How the Agent Works


The Legal, Regulatory, Trade & Compliance (LRTC) Analysis Agent follows a structured end-to-end classification workflow to ensure that all trade compliance evaluations are performed consistently, accurately, and using approved authoritative regulatory sources. When a trade classification request is submitted, the system creates a new classification record, generates a unique request identifier, associates the request with the originating system and product, and records the processing start time for traceability and audit purposes.

The agent then assembles a complete Classification Request Context Package using validated trade classification request files, Commodity Classification Worksheet entries, Decision Matrix responses, and supporting documentation linked to the applicable JDE Part Number. Supporting documents are processed using configured document processors to extract relevant content for validation and analysis. The system performs a multi-stage intake validation process that includes schema validation, document readability checks, document content verification, and validation result consolidation. Requests that fail validation or contain borderline conditions are halted to prevent inaccurate or unsupported classifications from progressing further. Borderline conditions are submissions that pass basic intake checks but contain issues that prevent reliable classification. These include: a required supporting document that is missing or only partially readable, a document whose contents do not align with its declared document type, a supporting document that cannot be clearly associated with the JDE Part Number referenced in the request, incomplete or inconsistent Decision Matrix responses, a mismatch between the product category declared by the user and the category identified by the agent, and inconsistencies between the information in the request fields and the information in the supporting documents. When any of these conditions is detected, the agent halts processing and returns a validation result identifying the issue so the user can correct and resubmit the request.

Once intake validation is successfully completed, the agent identifies the applicable product category and validates it against the category declared by the user. It then applies category-specific Decision Matrix logic and evaluates candidate Harmonized Tariff Schedule (HTS) headings using approved authoritative regulatory sources. The agent sequentially applies General Rules of Interpretation (GIR) logic until a classification is determined. If a classification cannot be resolved through the defined GIR process, the request is flagged for review by the LRTC team.

After classification is resolved, the agent evaluates applicable duty rates, regulatory considerations, and trade indicators before generating a detailed classification rationale. The system then assembles the Advisory Record and generates a downloadable determination output that is presented to users through the approved interface by the COS Team – Frontend. Throughout the process, the agent records all operational and decision events, maintains a complete assessment history, and stores the final determination record to support traceability, auditability, and future reference.

Agent Input


For the agent to process a request, the following must be present. Where any of these are missing, the agent halts processing and returns a validation failure rather than proceeding to classification: