This user guide provides an overview of the Virtual Mailroom (VMR), covering the user interface, key benefits and capabilities, the Human‑in‑the‑Loop (HITL) concept, document publishing, and best practices for maximising the value of your mail house solution. Virtual Mailroom (VMR) is an AI‑powered document intake and processing solution that automates how documents are received, classified, reviewed, and published, using advanced features such as automated document classification and data extraction to reduce manual handling while improving accuracy, consistency, and auditability. By streamlining end‑to‑end document workflows, VMR helps ensure fund administration processes are efficient, reliable, and future‑ready.

• Overview of the User Interface
• Key Benefits and Capabilities
• Common Use Cases
• Human-in-the-Loop (HITL) Concept
• Document Publishing
• What's Next?

Overview of the User Interface

The VMR dashboard provides a single, intuitive view of all documents as they move through the processing lifecycle. Access to VMR is available directly from the Class business menu using single sign‑on (SSO).

Key areas of the interface include:

• Documents Processing Queue
  Displays documents currently being loaded & analysed by AI before entering the automation pipeline.
• Documents Classified
  Shows documents that have been successfully classified as ready for publishing without further human intervention.
• Human‑in‑the‑Loop (HITL)
  Lists documents flagged with errors or concerns that require human review due to confidence thresholds, partial automation, or data validation requirements.
• Published Documents
  Provides a complete audit trail showing where documents were published (Class DMS or SharePoint), including who reviewed and published them.

Key Benefits and Capabilities

Virtual Mailroom delivers value across document handling, compliance, and operational efficiency.

Key benefits include:

• Automated document classification across a growing list of supported document types
• Data extraction from unstructured documents, converting content into structured metadata
• Reduced manual effort through automation and bulk processing
• Improved accuracy via AI confidence scoring and human validation
• Audit traceability with full history of AI actions and human overrides
• Secure handling of sensitive data, including automated TFN redaction
• Consistent document naming and tagging for easier retrieval and reporting

These capabilities are continuously improved through AI model retraining and sampling.

Common Use Cases

VMR is commonly used to automate the handling of high‑volume documents such as:

• SMSF financial statements and tax returns
• Investment statements and confirmations
• Member‑related documents (e.g. insurance, contributions, pensions, and benefit payments)
• Deeds, trustee declarations, and compliance documents

Once documents are ingested, VMR automatically:

1. Identifies the document type
2. Maps the document to the correct entity
3. Extracts key data points, including entity name, investment name, investment code, member name, document date, financial year, and signature detection (where applicable)
4. Converts extracted unstructured data into structured data and system tags for publishing
5. Uses the structured data to create multiple tag variations, enabling easy sorting and retrieval when documents are exported to Class DMS or SharePoint

This significantly reduces the need for manual document sorting and tagging.

Human-in-the-Loop (HITL) Concept

While VMR automates most document processing, Human‑in‑the‑Loop (HITL) ensures accuracy and trust where AI confidence is lower or where documents are only partially automated.

HITL allows users to:

• Review AI classifications and extracted data
• Correct or complete missing mappings
• Validate document readiness before publishing
• Provide feedback that contributes to ongoing AI learning

Documents that meet AI confidence thresholds can be published without any manual intervention, while others are surfaced in HITL and flagged as Errors or Concerns for quick and efficient review. This balanced approach reduces risk while maximising the benefits of automation.

Document Publishing

Once reviewed, documents can be published directly from VMR to:

• Class DMS – for Class‑administered funds, including document files and metadata
• SharePoint – for non‑Class entities, with structured folders, document files and metadata

Publishing can be done individually or in bulk.

What's Next?

Document Mapping and Automation Statuses in VMR