I have been working on enterprise AP automation programs for 20 years. I remember working with a large insurance company several years ago that painstakingly implemented a source-to-pay suite across all spend categories and regions. After 5 years of transformation — a massive change management effort to onboard thousands of suppliers on a network —, they still only managed to capture 55% of spend on POs and only 35% of invoice volume electronically. Today, they still have a large AP department chasing suppliers and requesters to process tens of thousands of paper and PDF invoices. They still manually figure out missing POs, 3-way matching, and determine the spend category, GL code, and cost center.
This company’s situation is, unfortunately, more the norm than the exception. I’ve seen the same challenges in hundreds of other enterprises: the long tail of requesters who won’t change behavior to buy on POs, and suppliers who won’t change the way they invoice, is a very real and complex issue that traditional automation solutions just can’t solve.
Over the course of my career, I’ve been lucky to witness and participate in the evolution of new technologies for finance, like ERP, finance analytics, P2P, T&E, OCR, and RPA, to address these challenges. The focus of all those waves of technologies was, and remains, to help users perform tasks, make decisions, and act faster with less manual effort.
While all of these have made a difference, users are still primarily responsible for reviewing, interpreting, and choosing the best course of action for all financial transactions.
Keeping users at the center adds human friction, errors, time, and cost to financial processes. It limits how much transformation is possible. And it’s the reason finance automation alone does not and can not deliver dramatic cost reduction, lean processing, agility, and improved customer experience.
Take the example of invoice processing: invoice processing costs and cycle times remain high and haven’t budged from the top two challenges cited by AP leaders year after year.
But now the world is in the age of AI, and it’s finally coming to the finance back office. AI offers a fundamentally different approach. Unlike the technologies of the past, AI can deliver truly autonomous invoice processing where AI software makes the decisions — extraction, validation, coding, matching, and invoice approval, rejection, or request for expert review.
By moving users to the edge of the process — toward the handling of exceptions, designing improvements, and monitoring performance, and away from basic processing — AI overcomes the fundamental limitations of previous automation technologies. That’s why I believe AI represents another inflection point.
To understand why the solutions of the past haven’t fully addressed the challenges for the AP team, let’s look a bit deeper at what these teams need.
To unlock value from inefficient and predominantly manual finance back-office operations, employees and suppliers need to change the way they buy and invoice for goods and services. Even the best finance automation software is not very effective if users don’t adopt it.
A distracted workforce — still adjusting to the shifting realities and burdens of “work from home” and “school from home” — is even harder to bring along on a change management journey.
According to research by Ardent Partners, nearly 97% of AP organizations have been affected by the coronavirus crisis. Most aren’t unable to manage paper processes remotely, so staff, classified as ‘“essential workers,” have been going into the office at personal risk. Digital AP teams have moved more easily to working from home, but even they are unable to easily process all invoices. The challenges of operating during the pandemic have only compounded on top of the challenges they faced before it began. The AP invoice process has been and continues to be highly manual, inaccurate, and burdensome for most AP teams.
Coming out of the covid-19 induced pandemic, most CFOs do not want to go back to the pre-pandemic manual processes and are striving for the maximum level of automation possible, giving a strong tail-wind for Autonomous FInance and AP programs.
AI-led Autonomous Accounts Payable is key to supercharging finance transformation
If the high manual effort at every step of the invoice process is the root cause of AP processing challenges, the solution can’t just be about assisting in human decision making and automation of the minutiae. Fifty years of work on this have proven so. The only way to transform the process and overcome its challenges is to rethink it as an AI-driven process, in which AP teams only need to address invoice exceptions while leaving the vast majority of invoices to be handled autonomously.
The great news is that technology has caught up to the problem. Just as the dream of autonomous driving is being realized with AI, accounts payable is also at the start of an AI-driven, autonomous revolution.
|AI-driven capabilities||Autonomous driving||Autonomous AP|
|Computer vision||Computer vision sees the surroundings e.g. roads, highways, people, objects||Computer vision extract information from invoices e.g. features, logos, numbers, text|
|Semantic understanding||Semantic understanding helps understand what is seen and distinguishes a person from a sign from a car from a lane marker, and understand their position, distance, direction, and relationships||Semantic understanding to make sense of the invoice, and apply the right taxonomies: What is being bought (spend category)? Who is buying it (cost center)? How should we account for it (prepaid, operating expense, etc), GL code|
|Automation & workflow||Translate the situation to action, for example by weaving left and stopping to avoid hitting someone||Autonomously process the invoice and assign next steps like a review or gather missing information|
|Human control- when needed||AI prompts the driver to take charge and control the steering wheel||AP workbench to review invoices where the AI isn’t 100% confident|
The benefits of autonomous AP applications are dramatic. Customers deploying these technologies are achieving:
- 60-80% end to end straight-through processing of non-electronic invoices, with 100% confidence
- reducing the cost of invoice processing
- reducing the number of manual touches and increase efficiency by up to 80%
- compressing invoice processing times to 2 days; capture early payment discounts
- generating savings of 0.1 – 0.3% across all third party spend annually
- driving down error rates
- bringing spend audit ahead of payment, and expanding the scope from spot checks to 100% of spend, reducing overpayments
- upskilling finance teams- greater focus on strategic activities; upskill AP staff to work with new-age technologies such as computer vision, AI, intelligent process automation
Additionally, these applications set the stage for a dramatic transformation, with much less pain, than alternatives. First, by reducing invoice cycle times and lowering barriers of supplier onboarding and invoicing changes they make suppliers and employees happier with the process. Second, they deliver value to finance leadership by improving working capital with early payment discounts and delivering significant savings — $8M+ a year for a typical $5B enterprise on invoice processing — with a rapid time to value, with ROI starting in 3-6 months.
The revolution that Autonomous AP offers finance leaders is real, and it is here. Companies are already seeing dramatic benefits from it.
To read more about the experiences and stories of enterprises who have successfully implemented Autonomous AP, and learn about the promise it can hold for your enterprise, take a tour of How autonomous invoice processing supercharges finance transformation – Guide to Autonomous AP.
Perception vs. Reality - The Use of Technology in Finance Departments
The Self-driving Finance Department - Fact or Fiction?