Vic.ai’s algorithm is currently reaching 97-99% accuracy. But, what does the accuracy rate really mean? It's a measure of how often the AI predicts something accurately. For us, that translates to 10-25 predictions per classification or line item in every invoice we process. Accuracy will naturally improve for two reasons: Our team constantly develops better algorithms and the algorithms see more data and are "trained" by receiving more corrections. This generates a positive feedback loop where accuracy will naturally increase over time.
Yes, Vic.ai can ingest and understand bills with literally hundreds of line items. It’s a unique strength of our solution. Line item coding might include posting to a dimension such as class, job, or location, splitting items across many General Ledger accounts, and even recognition and coding of VAT and other taxes. Both the header coding and the line item coding will be posted to your connected ERP.
Invoice processing automation, often known as RPA (Robotic Process Automation), is a rules-based process that extracts data from an invoice and uses it to populate all the required information into the accounts payable system. Usually these systems rely on memorization and templates to analyze invoices, a process that is easily broken when a new invoice format is presented. These systems don’t learn, and they require a lot of maintenance just to keep up with existing vendors, not to mention new ones.
In contrast, Vic.ai harnesses AI to run the accounting process. We’ve trained our algorithms on more than half a billion invoices, and we never need to resort to templates or memorization to extract the relevant data from an invoice. Our algorithms are 97-99% accurate at the header level from day one, and our system is constantly learning and improving. With our Autopilot technology, we can process invoices without the need to ever touch or review them because we are so certain of our predictions, both on the header and the line item level. This frees up AP teams from the monotonous work of reviewing (and fixing) an endless stream of invoices, so your company can focus on more strategic accounting tasks. Even better, the system feeds and cares for itself, and there is never any need to update a template or repair a broken process.
Vic.ai will detect duplicates across any document it ingests and will flag it as a duplicate using a red indicator.
Yes, Vic.ai integrates with a number of ERP systems by leveraging our open API. We are continuously adding new integrations, so if you don’t see your current ERP listed on our integrations page, please reach out to us here.
Once an invoice has been ingested, Vic.ai extracts the relevant data and automatically matches the invoice with the purchase order to verify that it is correct. The system will link each cost line on the invoice with the corresponding PO line using three-way matching. If the PO, goods receipt note, and invoice match, the system can process it automatically without the need for human revision.
Three-way matching is the process of matching PO, goods receipt notes, and invoices before issuing a payment of goods delivered. The three-way matching verifies that an invoice is correct by matching it with relevant documents that prove that the business ordered the goods or services and the business received the goods or services indicated in the invoice.
Autonomous approval flows
Your approval flows are generated based upon the flows and triggers created by your Vic.ai Organization Admin. Our approval system is extremely flexible and easy to use. An approval can be launched based on nearly any condition or set of conditions, including GL account, vendor, or amount, and dimensions such as location, class, and much more.
From the moment the approval flow is initiated, every comment, approval, and rejection is tagged to a user with a timestamp. The approval audit trail can even be posted along with the invoice to your ERP.
Vic.ai analytics offers customizable dashboards to provide insights on the data points that are important to your business. The dashboards include metrics such as Average time spent by users per invoice, Total invoices posted, Average total accuracy, Average time spent per invoice, Percent of invoices processed via Autopilot, Total accuracy per vendor, Average # of touches per invoice per user, and many more.
We offer multiple ways to access Vic.ai Analytics, through these different tiers:
- Standard Analytics - includes 5 standard metrics including: Average time spent by users per invoice, Total invoices posted, Average total accuracy, No touch vs touch invoices, and Total accuracy per vendor
- Advanced Analytics- Standard 5 tiles + a selection of 5 additional tiles from a designated list of previously curated tiles
- Custom Analytics - Login access is granted to Vic.ai Analytics with the ability to have unlimited tiles and a designated Vic.ai support partner
Invoices can be uploaded to the Vic.ai systems through multiple methods including direct file upload (mobile and desktop) and email ingestions.
Please contact our sales department to determine the best solutions for your company.
We can move as quickly as your organization wants to, and we offer both free administration and end-user training to help you get prepared. Optionally, Vic.ai can also train our AI model in advance of your launch by performing historical data training, which usually takes about 2 weeks.
The topic of languages can pertain both to the admin UI and to the processing of invoices that may arrive in different languages, different date formats, different currencies, and various methods of punctuating numbers (for example "," vs "."). Our current capabilities are outlined below:
- Invoice languages: Any Latin alphabet language is supported.
- Web app languages: Our UX, where admins, approvers, and reviewers analyze invoices, currently supports English, Swedish, and Norwegian.
As of February 2022, the Vic platform supports PDF, Doc and Docx, image files like .jpg, .gif, and .tif, and some electronic formats.