Core concepts

Understanding these key concepts will help you get the most out of Moby.

Project

Your engagement folder — contains Workspace, Collection, and all jobs for a client engagement. Each project is isolated and can have different team members with different access levels.

Every project must have at least one Manager. Managers can add/remove members, configure settings, and have full control over the project.

Workspace

Your working area for documents you're actively using in tests and extractions. Organize files into folders, preview documents, and select what to include in jobs.

Collection

Your PBC (Provided by Client) tracker. Create a document request list, monitor what's been received, and follow up on missing items. Syncs with SharePoint to automatically track incoming documents.

Models

Models are AI workflows — reusable templates that tell Moby what to do. They're created by your firm or yourself and can be shared across projects and clients.

Think of models as recipes: once you build a good one, you can use it over and over without starting from scratch.

Types of models

Type	What it does	Example
Extraction model	Defines what data to pull from documents	Invoice fields: vendor, date, amount, line items
Test of Details model	End-to-end audit test — match documents to transactions, extract data, and analyze results	Verify purchases: match invoices to GL, extract amounts, check for discrepancies
Analysis model	Defines data analysis workflows	Recalculate totals, flag anomalies

Creating and sharing models

• Generate with AI — Let Moby create a model from a prompt, a workpaper, or a set of sample documents
• Build from scratch — Define fields manually with descriptions
• Duplicate and adapt — Copy an existing model and tweak it for a new client

Model visibility

Visibility	Who can use it
Private	Only you
Client	Anyone working on that client
Firm-wide	Anyone in your organization

Tips for building great models

• Use descriptive field names — "Invoice Total Amount" beats "Amount"
• Add field descriptions — Explain where the field typically appears and common variations
• Test on a few documents first — Run on 3-5 samples before a large batch
• Iterate — Refine descriptions based on extraction results

Job

A single run of extraction, TOD, or analysis. Each job produces results you can review and export. Jobs are immutable — once completed, results are preserved for audit trail.

Matching

The process of linking supporting documents to transactions in a Test of Details. Moby uses AI to automatically match documents based on:

• Invoice numbers
• Dates
• Amounts
• Vendor/customer names
• Other identifying information

Grounding

Visual proof showing exactly where Moby found a value in the source document. When you click on an extracted value, Moby highlights the exact location (bounding box) in the PDF where that value was found.

This is critical for audit documentation — you can always trace back to the source.

Reference document

The Excel file containing transactions you're testing in a Test of Details. Your reference document is the list of items (population) you want to verify against supporting documents.