Onboarding Non-API accounts

How to submit historical data to train and connect Non-API accounts

Ryan Goetz avatar
Written by Ryan Goetz
Updated over a week ago

Historical Data Training outside of directly connected Vic Accounts:

Our team at Vic.ai can set up any client that cannot connect through API, we call these Non-API clients, for reference, typical API clients are bill.com, Intacct or QBO.

For a Non-API client, completing processing within Vic.ai would include exporting a download of data, that would allow you to easily re-upload that export into your accounting system.

In order to do historical data model training on these accounts, we would need the following:

A. Invoice copies

Ideally we receive several months worth of historical invoice copies (pdf or other format) so we can train our system on this data. The result will be high accuracy of automatic data capturing once you are on VIC's system.

  • Required is a minimum number of invoices (500 but the more the better)

  • File requirements: PDF, jpeg (or other images). One file per invoice

  • To train the data on all submitted invoices, a consistent naming convention is helpful so we can match the invoice file to the invoice data inside the submitted csv.
    If you have internal reference id’s an example naming convention that would work: <qbo_invoice number_reference id.pdf>

B. Transactions report (.csv file)

We require an export from your accounting system covering the same period of data history so that we can map invoice line items to data that the AI will extract automatically (ref numbers, due dates, issue dates, gl code vendor …). Ideally this report is provided in the format laid out in this spreadsheet.

A requirement of the transaction report is a link in the .csv file between filename (pdf invoice copies) and transaction.

In other words, the AI model requires the actual invoice copies + the information of expense GL codings. There has to be a way to determine which expense line item in the csv export matches which invoice copy.

C. Going live with new accounts in 2 week batches

Our team collects new account data and performs AI model training based on the historical data available in 2 week batches. Let our team know if you would like to submit data for a new or existing account and we can let you know the cutoff date for submitting the data and correlating go-live date.

Our team is happy to provide a shared folder link, at which you can upload the historical data.

If you wish to onboard a new client with us, please feel free to submit an onboarding request by filling out this form, and a member of our team will be in touch for next steps!

Or if you have further questions, don’t hesitate to reach out to our support team in the chat bubble or by emailing support@vic.ai.

Did this answer your question?