One of the most complex parts of learning Vic.ai is interacting with expense line data.
The way that the AI extracts line items, as well as how to interact with the Gl accounts and dimensions on them is explained below.
Groupings & Line Items
Groupings are a collection of expense lines that all share the same coding for GL accounts and any dimensions. A grouping can consist of one or many line items.
There can be multiple groupings per invoice
Groupings post to the connected system, not the individual line items except for rare configurations that specifically request this.
If you edit coding at the grouping level, changes will cascade to all expense line underneath the grouping.
To update the GL account or dimension, choose the correct value from the dropdown list. As you type numbers or letters, the options shown to you will narrow down!
Line Dimension Dependencies
If configured, when selecting dimensions, only valid options will appear based upon data already selected.
For example, once you select a Department, only the Locations (or other dependent dimensions) that are valid for that Department will appear — reflecting the rules configured in your ERP.
Dependent dimension lists filter in real time as you code
Invalid combinations are blocked before saving — invalid options simply don't appear in the list
If no controlling value is set yet, the dependent dimension shows the full unfiltered list
If you change or clear a controlling dimension value and your previously selected dependent value is no longer valid, Vic.ai will surface the conflict and prompt you to resolve it (or clear it automatically, depending on your company's configuration)
Multi-step dependency chains are supported — for example, Dimension A controlling B, which controls C
AI predictions respect dependency rules when suggesting line values, so suggestions will only include valid combinations given what's already set on the line
This requires a special integration from your ERP, if you're interested in enabling this, please reach out to your Vic.ai representative.
AI Prediction of Expense Lines
This is the most complicated calculation performed by the AI. It will predict the total amount of an invoice, as well as all of the potential expense lines. The AI then attempts to make combinations of the expense lines that sum up to the predicted total amount.
The predicted line items are difficult for invoices with excess data that look:
Look like expense lines
Have repeating subtotal and total lines.
Common offenders are utility and telecom bills. Patience while the AI learns from numerous examples will be rewarded!
Line Item: Insert and Delete
This should be used sparingly, and only if the predicted amount and line items are incorrect.
Insert a line item: Use the keyboard shortcut letter “i” or click on the plus sign.
Delete a line item: Use the keyboard shortcut letter “d” or click the red “X.”
Inserting a line item will copy the coding of the selected line item
AI Learning: Splits
One of the defining features of Vic.ai is that the AI will learn how your company codes their books to your unique Chart of Accounts and Dimensions.
To facilitate this, update the coding for groupings, or even individual expense lines. Doing this repeatedly reinforces the learnings for the AI, and it will begin to shift the predictions and confidence scores for the predicted expense lines.
If you're looking to perform Splits within Vic.ai by splitting an existing line item, the AI won't learn this as the information does not reside on the Invoice document. Instead, hopefully your connected system supports Allocations, and the split can be performed within that connected system.
What text-only fields posts to a connected system?
Generally, the “Bill Description” at the header level and the “Item Description” at the line item level will feed over to your connected system.
The scanned “Item Text (On Invoice)” does not feed into the connected system.
