Increase accuracy and precision of your invoice processing with updates to’s AI model

Jessica Sadow
Jessica Sadow
Director of Product Marketing

Discover how updates to's AI model further increase invoice processing accuracy and efficiency. Learn what makes our AI unique and how it adds value to your Accounts Payable (AP) workflow.

August 31, 2023

4 min read

Increase accuracy and precision of your invoice processing with updates to’s AI model

We are excited to announce enhancements to our AI model, giving our customers more accurate and nimble invoice processing.  AI has been at the core of our platform since our founding and we are constantly looking for ways to push the envelope on what is possible in autonomous finance on behalf of our customers. 

While the headlines about AI are attention-grabbing and there’s a lot of interest in AI, we’d like to break down what AI means for organizations and how we think about it as a company. 

When thinking about real-world applications of AI, it boils down to two key components:

  1. How is AI helpful? This is about understanding what value it brings to the table and how well it works with its human counterparts to enhance work and increase value/productivity.
  2. What is the AI made of? By this, we mean what data was used to train it and how it learns.

How is’s AI helpful? 

For years, financial institutions have been streamlining their operations through rule-based tools such as RPA. However, mere "automation" is no longer enough. Tools based on templates are inflexible, unwieldy, and expensive to implement and maintain. Because any changes to an invoice mean manually updating templates, it still requires significant hands-on work. This approach becomes impractical for organizations that handle a high volume of accounting transactions each month. 

AI, on the other hand, goes beyond conventional automation because it does not rely on rules or templates. When integrated into finance and accounting processes, AI can deliver enhanced efficiency, precision, and autonomy. Through its adaptive and logic-driven nature, AI aids in the independent categorization of invoices, facilitates approval management, and expedites the matching of purchase orders, surpassing human capabilities in accuracy and speed. Comprehensive data streams into dashboards and financial reports, empowering finance leaders to extract valuable insights from transactions in real time, thereby expediting informed decision-making.

The results of truly autonomous invoice processing speak for themselves. At our customers have achieved 99% accuracy, 5x productivity, 80% reduction in invoice processing time, and realized ROI within seven months. 

And with the updates to our model, we expect to see these metrics continue to improve for our customers. In particular, customers will benefit from:

  1. Improved accuracy: The latest and greatest in deep learning technology, combined with the quality and quantity of our data means high precision and accuracy of our predictions.
  2. Flexibility at scale: The new model makes it easier for our engineering teams to quickly add in new fields and paves the way for international expansion. 

What are the ingredients for’s AI?

We are pioneers in the use of autonomy and intelligence to transform accounting and financial processes. While we’ve been building AI since 2016, AI has rapidly developed in the past 2 years and the updates to our model reflect that. 

Our new model takes an updated approach to applying the LLM to invoice processing by incorporating behavioral inputs, a more comprehensive analysis of the document image, and a more scalable and flexible solution. 

By utilizing the most advanced LLM technology, clients will experience the enhanced agility of state-of-the-art AI. On top of this, our proprietary AI model is trained on layers of data. Combined with its continuous learning, it delivers autonomous invoice processing with speed and accuracy to improve productivity, decision-making, and ROI.

To achieve this level of financial expertise, our AI model incorporates data from the broad base of its training materials, to the tailored learning required to deliver results to each customer:

  1. Deep training: Out of the box, our model trained on over half a billion invoices.
  2. Extensive on-the-job learning: Since launch, our model has trained on hundreds of millions of invoices with our customer deployments.
  3. Studious: As part of the customer onboarding process, our model learns through customer historical data so that it tailors its processing to their specific needs. It not only learns from the data on the invoices but also from my user changes made on the historical invoices.
  4. Adaptable: And it doesn’t stop there. As customers make adjustments - whether to the data, adding new vendors, or changes in the invoice format, our model gets better each and every day.

While our old model only could surface data that was directly on the invoice, with the latest LLM technology, our updated model generates text based on past customer actions. The model can learn from individual inputs and make predictions even if they are not present in the document image. This means the data surfaced is tailored to what the customer expects. 

This new model is better for customers as it can generate predictions based on learned user behavior, even if the due date or other details aren't directly written on the invoice. For example, if a customer creates their invoice numbers in a unique way, by combining the number '0', the date on the invoice, the rounded total amount from the invoice, and the last four digits of the account number on the invoice, our new model, can understand this and consistently guess the invoice number correctly. 

We are initially rolling this out for all AI predictions made using invoice header data but we will expand it across other data sets over time. 

AI gets better with experience 

The future of autonomous finance holds the promise of seamlessly integrated AI systems that autonomously manage and optimize financial processes, from intricate transactions to strategic decision-making. This evolution is set to revolutionize traditional financial paradigms, offering unparalleled efficiency, accuracy, and adaptability in navigating complex economic landscapes.

Here at, we are excited that by employing the latest in LLM technology, customers will benefit from the increased nimbleness of cutting-edge AI combined with the strong foundation of our deep financial expertise. 

We are committed to making sure our customers are at the forefront of these advancements and providing best-in-class autonomous AP. 

Vic AI Demo See AI Accounting in action


Sign up to receive the latest news, updates and valuable accounting and finance resources conveniently in your inbox.