Why computer vision AI is the next BIG thing in accounting

Molly LaMantia

Molly LaMantia

Marketing Manager

Learn why AI is a better alternative than RPA with ICR to automate invoice processing, and how they differ. Increase the share of no touch invoices within weeks.

August 15, 2022

4 min read

Why computer vision AI is the next BIG thing in accounting

Accounting praxes and technology adoption have steadily evolved. A few decades ago, it would have been common for accountants to manually record and reconcile company books. Manual bookkeeping was, and still is, a tedious and time-consuming task for companies and in-house accountants. This, in turn, leads companies to offshore (or outsource) their tedious accounting tasks to lower their internal costs — even during high seasons for accounting. 

Role of AI in accounting 

Fast-forward ahead a few decades; there’s a shift. With the invention of rules-based automation (RPA) and intelligent character recognition (ICR), companies opted to bring data entry home to assist with manual work. And now, with AI, accounting can finally be optimized to minimize human assistance and empower the accountant to engage with more value-added activities and enjoy more job satisfaction. 

Unfortunately, many companies are still using manual processes since they haven’t made the switch to AI. In this article, we’ll break down why AI is a better alternative than RPA with ICR. 

Why is AI a better alternative than RPA with ICR? 

AI-driven accounting is agnostic to document templates, and can read and process all types of documents, whether it’s invoices, purchase orders, contracts, or receipts. A true AI-driven model is fast to roll out and can quickly establish and maintain a high accuracy despite a change in suppliers or volume. The latter is vital when volume increases - with AI in place, no extra manpower isn’t needed. 

When implementing an AI-driven solution, the AI trains on historical data sets, like documents, charts of accounts, etc. The model training configures and stress tests customer and their business as a part of the onboarding process. A good model learns and calibrates with time, is accurate in real-time, and provides measurable business and operational insights.

The downside of RPA with ICR

On the other hand, an RPA with ICR model relies on template training of documents with a chart of accounts mapped via RPA. It implies that the model must be retrained to support the new documents as the suppliers or transactional volume changes. This makes it a less reliable real-time solution requiring endless human intervention. 

It’s never once and done but requires a continuous engagement. 

In praxis, it means that you either need to build your IT or automation staff or have a contract for an external party to maintain the solution to support your automation continuously. While this may alleviate the accountant’s tasks, it is no scalable or cost-effective solution.

Example of AI accounting system in action

When choosing an AI system, one crucial consideration is the need for human involvement in invoice processing. As an example, the graph below shows the total number of invoices that were touched vs. not touched by a human reviewer throughout 4-months with the platform: 

Autonomous AP Processing Graph

The AI learns and gets smarter over time with feedback and iteration from the AP team. The graph shows a steady progression of how AI can learn from your personnel to make better decisions over time — on its own. 

Unlike a rules-based technology with ICR, an IT specialist or contractor would need to update the rules or template to reflect invoicing operations. This means every new invoice format from a new or current vendor needs a new workflow created for the ICR to read the document. At the same time, our AI system learns from historical mistakes to relieve accountants from processing the documents and transforming them into reviewers. This also frees up IT and automation staff capacity. 

“Why should I care about computer vision AI as a CFO?”

So, why should CFOs care about AI in its relation to accounting? Quite simply, it is the fiduciary responsibility of the CFO to close the books and report on financial status with accurate and timely data. Additionally, when enterprise-level finance departments break silos between AP, AR, FP&A, etc., they can seamlessly combine efforts by integrating their technology systems and processes. Everyone wins. 

A CFO should not only be equipped with business and operational insights but should also be in a position to take timely action. A real-time accounting platform can:

  • Increase efficiencies by relieving the accounting team of mundane tasks such as data entry or cost center allocation and, in turn, attract and retain top talent.
  • Optimize potential with supplier discounts by minimizing time spent on approvals from 15-20 days (industry average) to instantaneous.
  • Avoid fraudulent payments. A typical organization loses 5% of its revenue to fraud every year. With a clean audit log and duplicate detection, you can save your company thousands of dollars.
  • Minimize time spent in support and maintenance. 
  • Enable course correction of the business in real-time with invaluable insights on cost, spending, efficiencies, bottlenecks, operational cash, etc. Waiting to get these insights at month-end or quarter-end is too late. 

Transform your business with AP autonomy and intelligence

As you continue your search for a computer vision AI system, there are many things to consider — but most importantly, you must consider how the technology will integrate with the teach-stack your teams are already using. 

Every technology has its strengths, and ours is that we are integration-friendly! At Vic. ai, we love our ERP friends. Our system integrates seamlessly with your current system, and you can be up and running within four weeks. 

Additionally, our accounting-focused system: 

  • It doesn’t need template training or setting up rules.
  • It is purely AI-based — unlike competitors who use a predominantly OCR/ICR system with limited AI features. We believe in 100% transparency!
  • It doesn’t require GL coding when workflows change — because we don’t use workflows. 


The bottom line is this. Investing in technology requires more than a spreadsheet ROI. It must be championed and adopted internally, provide a good user experience, and grow with the business.

Future-proof your finance department

See why thousands of CFOs globally put their trust in us. Schedule a demo to exceed your growth and bottom-line goals by accelerating digital transformation.

Want to see how Countsy, an industry leader in Accounting as a Service for venture-backed startups, such as Asana, Instagram, Intercom and Doordash spends 84% less time per invoice and has gained 2.3X more capacity per FTE with accounting AI? Click below to download the case study.

Be a Finance Pioneer

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