• Retrieving invoices or documents from customer or supplier portals
• Processing transactions
• Passing documents to other systems, e.g. sending invoices for approval or logging them in an ERP system

Artificial Intelligence (AI)
AI is the theory and development of computer systems to perform or enhance tasks normally requiring human intelligence. AI improves speed and accuracy within document processing by completing tasks such as:

• Perceiving and interpreting documents
• Matching PO invoices with corresponding PO lines and goods receipts
• Identifying anomalies and predicting outcomes

Machine Learning
Based on input from users, a solution such as Esker’s learns how to process new document formats and expand its knowledge base. The more it learns, the more the recognition rate increases and the system can process documents of the same format without any input from users (touchless processing).

Deep Learning
Taking machine learning one step further, deep learning is a neural network using algorithmic software to train itself to perform tasks based on a large set of data. Put simply, the software analyses data to find patterns in the data so it knows what to expect and what to do with a new document when it comes in. Esker’s own deep learning technology, Synergy, has been trained by the millions of documents processed by its solutions over more than a decade, allowing businesses to benefit from high recognition rates from day one.

The Benefits
A digital solution such as Esker uses a combination of these technologies to automate manual inefficiencies and low-value tasks in the P2P and O2C cycles. This frees up teams to focus on more value-adding tasks, increases efficiency and reduces cycle times, benefiting not only your business, but your customer and suppliers too. Businesses set their own pace, choosing to automate one process at a time (Procurement, Accounts Payable, Order Management or Accounts Receivable), or automating entire cycles to unite their operations across a single, integrated platform.

Find out more
If all this technology talk has you buzzing to update your systems, take a look at the following resources for a more in-depth look into the benefits of process automation for the procure-to-pay and order-to-cash cycles:

AI & RPA: The Technology Powering P2P Digital Transformation

New Technologies for Order Management – Increase Your Competitiveness

Jennifer Ball

As Marketing Co-ordinator for Esker UK, Jennifer manages Esker UK's marketing campaigns and events for S2P solutions. She has been part of the Esker family since 2019.

Read more insights from Jennifer Ball

Let’s dig into what they are, how they can help and where processes fall short.

Robotic Process Automation

Often seen as a stand-alone solution, RPA is a “robot” that is programmed to push or pull information from one place to another. RPA has strong benefits for simple business processes because it can eliminate repetitive, manual steps. In short, if you can draw a virtual line from one point of a process to the other, RPA may be a good fit for it. An example of this would be: logging into a portal, downloading a document, then pushing the document to a folder for someone else to review. Sometimes, if the data is perfect, that data can be pushed reliably into a database/system of record after it has been downloaded by the RPA/robot.

There are even some cases where a robot can be programmed to read the data it’s pulling and make a simple determination about where that information goes. These are often times referred to as “if statements.” “If statements” are variables that determine where information should go but not necessarily validate if that information is correct. Changes occur on either end of the process where the RPA identifies any weakness. What happens if login information for a portal expires or the process changes on the other end? Consider the downstream effects of these changes and ask yourself, “Are there some areas in my business that would benefit from a simple and effective automation process?”

Artificial Intelligence

AI is considered a technological “brain.” Often construed as a technology for extracting data, AI is really just a technology that assists in handling data and information. In other words, it’s a group of technologies and algorithms that help solutions make accurate decisions based on trends and past decisions. But not all AI is the same — some AI technologies are complex and others are rather simple. Overall, if a technology is helping to make a decision — without a user forcing or programming it to — it can be classified as artificial intelligence.

So, what is AI most useful for? Most experiences suggest that AI “decisioning” technologies are best used to cut down on time-consuming, manual tasks. AI mimics decisions that otherwise take additional time for humans to make. The ROI calculation is a simple equation. Here’s an example: if one step in a process takes 10 seconds to complete and is repeated 100 times per day, AI intervention can save 16.6 minutes a day on that single step. It’s easy to see how ROI can grow to be substantial as these seconds can turn into minutes (or even hours) of time savings. Complex processes seem to benefit the most from this technology.

Machine Learning

In its truest form, machine learning is a variation of robotics in which the solution is manually told what to do. These are forced or programmed decisions based on what a user recommends. Despite all the promises that technologies like RPA and AI make, there are still situations where specifically telling a solution how to handle a piece of an example is necessary. Think of this as the technology that will handle the “details” of a business process.

Let’s say, for instance, you have a customer that consistently changes a material number when placing an order. As a CSR, you know what the customer is trying to order, but the change happens far too often for the cognitive functions of AI to intervene. In this example, machine learning can step in and force the system to defer to the correct change — just as a human would do given a manual process.

Think about your business process and ask, “Is my process complex or is it simple? Does the business process have variables that require constant change? If changes aren’t corrected in a timely manner, will it negatively affect my customer or supplier relationships?” In answering these questions, you may be able to determine what technology is best for your business process.

When it comes to customers and suppliers, leveraging all of these technologies has proven to produce the best outcomes. Solutions tied to these areas of focus should be flexible and retain the ability to grow and change with a business. Customers and suppliers are arguably the lifelines within a business and should be treated with the utmost care.

Written by Chris Wadley, Business Development Manager for Esker

Esker UK

Unlocking Positive-Sum Growth with AI-Driven Business Solutions for P2P & O2C Cycles

Read more insights from Esker UK

Clément Sage, a machine-learning PhD student and engineer at Esker, will present his paper titled “Recurrent Neural Network Approach for Table Field Extraction in Business Documents” in the context of his PhD research at Université Claude Bernard Lyon 1 in France.

Efficiently extracting information from incoming business documents, such as orders and invoices, is crucial for companies that face massive daily document flows. These documents contain valuable information that companies need to retrieve for integration in their ERP system and structured archiving. However, automating data extraction is extremely challenging, especially when analysing table content to identify ordered or invoiced items, as the documents have complex and ambiguous structures.

To address this problem, Esker developed an end-to-end method for table fields’ extraction by bypassing
physical structure recognition. Esker proposes a generic approach based on deep learning, which enables order item identification on layouts not necessarily seen during training. Sage’s first-time recognition approach is of particular interest to researchers, as it is generic enough to easily be adapted for other document types and relies on little domain specific textual preprocessing. This universal and nonproprietary AI algorithm is an important discovery for the document community.

Esker successfully evaluated the effectiveness of this approach on real-world purchase orders to retrieve product numbers, quantities and unit prices from ordered items. Following positive results, Esker is now using this technology on its cloud-based platform. The company’s customers are already benefitting from improved data recognition on new documents and better automation rates.

Read full press release here.

Esker UK

Unlocking Positive-Sum Growth with AI-Driven Business Solutions for P2P & O2C Cycles

Read more insights from Esker UK

In an age of constant change the core objectives of credit management have remained unchanged for decades; maintain healthy cash flow, minimise bad debt and reduce exposure to risk.

However, in the eternal race to find better, faster and cheaper ways of working, businesses must constantly improve how they deliver these objectives. The arrival of ERP systems several years ago fuelled and supported the growth of financial systems to deliver such improvements, and it’s the case today with AI to take this to the next level.

There is much debate surrounding AI and it’s a subject that can certainly divide opinion. While some view development with excitement and optimism, others deem it a threat; a dangerous endeavour that will take over the world.

The fact is that we are witnessing a profound transformation that is affecting markets, companies, our world and our own personal lives. Technologists predict that AI will bring about the greatest step-change to the global economy since the last Industrial Revolution. Yet AI, machine learning and robotic automation are already affecting the way we live and do business, and to a far greater extent than we may realise. So, perhaps the future heralds an evolution, rather than an AI revolution?

A brief history of artificial intelligence
The phrase ‘Artificial Intelligence’ was first coined in the 1950’s by John McCarthy, when he held the first academic conference on the subject. But interest in AI and robotics began well before then. Examples of independent-thinking artificial beings have been staples of pop culture for decades — from Mary Shelley’s Frankenstein (1818) to Stanley Kubrick’s 2001: A Space Odyssey (1968) and James Cameron’s The Terminator (1984).

In the 1960’s and 70’s there were steady improvements to AI capabilities. Computers became cheaper, more accessible and had greater computational power (computer storage and processing speed). There were high hopes for AI; it’s commonly cited that in 1970 Marvin Minsky told Life magazine that ‘In from three to eight years we will have a machine with the general intelligence of an average human being’.

This proved optimistic as, nearly half a decade later, human level machine intelligence is still to be realised. But since then machine-learning algorithms have significantly advanced, as the memory and speed of modern computers can retain and process vast amounts of data.

Varying degrees of intelligence
So, whilst the question of whether a computer can truly think remains unknown, a computer’s ability to process logic is undeniable. One technology with roots in the application of programmable logic is RPA.

RPA drives innovation within financial processes by automating repetitive, manually-intensive tasks and workflows. It is particularly effective in automating data entry and processing of invoices, due to its ability to quickly recognise and extract data within structured documents.

Since its introduction in the early 2000’s the RPA market has developed rapidly as businesses explore new ways to apply the technology to processes and workflows in a bid to optimise business performance. By 2022 Gartner research estimates that 85% of large companies will have deployed some form of RPA, and spending is on pace to reach $2.4bn.

RPA isn’t a new concept though, at its core is a pre-defined set of algorithms to allow the automation of high-volume, structured tasks with the goal of increasing efficiency. Exceptions can be managed by ‘teaching’ the software new ways of handling specific circumstances, but it has no in-built intelligence, relying on human input to tell it what action to take.

AI that utilises deep learning is a more complex technology than RPA, but also more powerful in terms of understanding process complexities and discovering the optimal solution required.

Deep learning is learning based on a multi-layered neural network as opposed to task-specific algorithms. For example, with deep learning you can train computers to build algorithms that know how to deal with complex issues or make decisions with an expected outcome in a given situation. It’s AI that makes it possible to develop autonomous cars, automatically detect medical anomalies, or even win a game of chess against a human champion.

Solutions being developed within the area of credit management such as Esker, a worldwide leader in AI-driven process automation and pioneer in cloud computing, will see increasingly improved processes within an organisation. For example, using data available on a specific company (such as that sourced from credit bureaux) and then analysing this in relation to the behaviour of how a specific company has paid its invoices previously (over a set time frame and possibly using other departmental data within the company such as orders or claims) can prove to be very beneficial. Basically, this would allow the most appropriate credit limit to be proposed and recalculated automatically in the future through a deeper learning of the information.

Deep learning also allows Esker’s platform to sort messages received from multiple channels based on the nature of the document (e.g. invoice, order form, spam, etc.) or the language used. Plus, it can open a document to check if it contains one or more invoices and send them to the correct approval workflow – all such non-value-added tasks previously undertaken by finance professionals.
AI and deep learning today are located on the perimeter between research and applications, however their diffusion into the real world will be rapid and profoundly change the nature of back office functions.

Technology’s impact on financial operations
AI automation can transform credit management, increasing the speed and accuracy in which routine tasks can be undertaken allowing real-time visibility with customisable dashboards and built-in KPIs. By freeing up people from low value and repetitive tasks, RPA and AI can lead to increased employee empowerment to focus on strategic accounts to improve customer relationships or generate high-level reporting for more precise decision making. Furthering professional development can also be gained through the opportunity to access a broader skill set.

Credit management therefore gradually shifts from an operational, task-oriented function to one of analysis, management and fraud control. Technology allows financial operations to not only measure performance in real time, but also identify problems or opportunities as they arise and thus become a key player and partner in a company’s strategic development.

Bringing together human and artificial intelligence
Even with the many benefits that RPA and AI can bring, technology does not replace the need for people, because it cannot replicate what are truly human skills. It cannot think creatively or imagine solutions by itself, it cannot apply social or emotional intelligence to a situation.

So, whilst AI and its role in the workplace continues to evolve, the alarm bells are fading as a more balanced view emerges. The robots are coming, but they will bring neither an apocalypse nor a utopia. At Esker we believe intelligent automation is the combination of AI and human intellect – the melding of the best the two have to offer.

About Esker
Esker is a worldwide leader in cloud-based document process automation software, helping financial and customer service departments digitally transform their order-to-cash (O2C) and purchase-to-pay (P2P) cycles. Used by more than 6,000 companies worldwide, Esker’s solutions incorporate technologies like artificial intelligence (AI) to drive increased productivity, enhanced visibility, reduced fraud risk, and improved collaboration with customers, suppliers and internally. Esker operates in North America, Latin America, Europe and Asia Pacific with global headquarters in Lyon, France, and U.S. headquarters in Madison, Wisconsin. For more information on Esker and its solutions, visit https://www.esker.co.uk. Follow Esker on LinkedIn at Esker – Northern Europe, or on Twitter at @EskerNEurope and join the conversation on the Esker blog.

Read full press release here.

Alistair Nicholas

Alistair is the Managing Director of Esker Northern Europe

Read more insights from Alistair Nicholas

To raise even more money, I donated some paintings for their summer auction. Hopefully people will be kind enough to part with their cash generously!

Speaking as an animal lover and a vegan, I was struck by how hard animals, such as donkeys, work for us. As living beings, like us, they get old, tired and need to take it easy. Happily, such considerations do not apply when deploying a Robotic Process Automation (RPA) which is designed to work 24/7 if required.

RPA is the buzz word for software that can be easily programmed to do routine, repetitive human tasks, quickly, accurately and tirelessly. Relying on structured data, RPA automates workflows or clerical processes by emulating human interaction within a graphical user interface (GUI) with great benefits. The benefits include consistently swift, accurate data management at a much lower cost than manual processing of any documents, typically sales orders and supplier invoices. Manual processes are streamlined to facilitate business security and increase scalability.

What does RPA do for your most important resource, i.e. your employees? Certainly staff are empowered to be more productive and professionally fulfilled because they are free of the mundane, repetitive and valueless tasks.

What of the other buzz word, Artificial Intelligence (AI)? AI is not to be confused with RPA. True, both deal with automation. However, a key difference is that RPA is not “self-learning” and only works with structured data. AI technologies, on the other hand, respond to changing environments and data and rewrite themselves.

What both RPA and AI have in common is that they automate manual processes and usher in great efficiency. Both technologies work 24/7 so you don’t need to. This is good news for staff, who are now able to work smarter rather than harder. Let RPA, and AI (with its machine learning and deep learning capabilities) take the strain of today’s a fast paced, IT driven environment. For me, that means more free time to serve our clients better and of course, paint more pictures and visit our adopted donkeys in their pasture lands!

Raj Sahota

As Internal Sales Manager at Esker, Raj looks after customers and prospective customers. She has been part of the Esker family since 2011.

Read more insights from Raj Sahota