A successful combination of human and artificial intelligence is rocket-charging invoice processing for Konica Minolta’s clients. But having found the sweet spot, it won’t stop at accounts payable, says Francis Thornhill, Marketing Director for Intelligent Information management and Managed IT Services.
In October 2019, Konica Minolta Business Solutions (UK) launched its smart invoice processing solution. A Cloud-based service, it aims to address and eliminate the two main troublesome areas of invoice processing – non-standardised invoice formats and the additional manual labour required to make up for accuracy issues created by last-gen automation.
Automation through artificial intelligence (AI) and machine learning doesn’t work well when the data is not standardisedor is unstructured; it’s the reason most automation solutions available on the market have limits and fail to deliver impactful, accurate solutions, says Konica Minolta. It claims to have addressed this with accurate extraction and capturing of data from a wide range of sources, re-entering the data itself, if necessary, in a format recognised by the software, as part of its service offer, which is delivering huge value to clients
The application of AI in tandem with human intelligence has, it says, combined to create amazing synergies, enabling its platform to reclaim the time an average accounts payable professional spends on so-called ‘automated’ transaction processing. And it is offering this as a service, on subscription, even to those not buying the invoice processing platform.
The Fintech Magazine spoke to Francis Thornhill, marketing director for intelligent information management and managed IT services at Konica Minolta, to get a better understanding of how it’s working smarter.
TFM: Who are your ideal clients in invoice processing – big multinationals or smaller businesses?
FT: We’re ideally placed to help companies that are processing 1,000 to tens of thousands of invoices a month. Basically, businesses of all sizes, including multinationals. Our platform and service is scalable and configurable to fit the largest customers – or, indeed, those that might not sign up to our platform. They could just take our capture service, for instance. But we also support smaller customers. Some businesses have very few invoices, but lots of line items, and they all have different delivery dates. All these present challenges when you get errors in data entry. We can handle all of that for our customers – even small businesses can leverage our Cloud service.
TFM: Why would a business want to outsource the core task of accounts payable?
FT: In the accounts payable profession, people mostly find themselves doing data entry or verification of invoices. There’s no standard for invoices; every one invoice is different, sometimes even handwritten. So, when you put them through the automation pipe, even if you are using AI, you start to get failures. What we’re trying to do is sort out the real heart of the issue, which is errors in capture. We do capture verification for our customers as a service so that they can focus on the real value-add within accounts payable. This way they don’t lose control of decisionmaking; the customer still makes the final decision, but right up until that point, it’s automated through our invoice processing solution.
TFM: What makes your capture service so different?
FT: We’ve developed a capture service where we do the verification and rekeying, if necessary, to improve automation levels, and address the real manual and repetitive burden that accounts payable teams are facing. Our core strength is in the capture side, making sure that the data, from whichever source, can be extracted. The data is captured, classified and then made ready for workflow automation. Key to any successful automaton project, whether they are finance related or any other, is the quality of data going into the workflow and automation. That is why the ability to capture, extract and classify data from any source is so important before the automation and workflow phase. Our platform and other solutions can use software robots that take data from web services and application programming interfaces (APIs). We deliver all this as a Cloud-based service, with a subscription so that accounts payable teams aren’t faced with a massive capital investment, which is one of the key barriers to transformation.
TFM: What is the motivation for businesses to automate accounts payable and invoicing?
FT: I think there’s a whole collection of challenges and concerns. Mostly, they realise that the existing way of doing things is too manual and it’s preventing them from making the progress they need in terms of efficiencies, visibility or control. The one or two per cent improvement each year is no longer good enough and blocking progress. So, they realise they’ve got to make a step change in the way they do things to achieve new levels of efficiencies, or even new services. They recognise that it’s not just a case of buying software, or the quick fix, but that they’ve got to do it differently – properly – and that means across people, process and technology.
TFM: How difficult is it to work with the banks, given their legacy systems and corporate culture?
FT: We actually enjoy working with the banks, because they’re really methodical and robust. We insist on using a discovery assessment before we start any project. The discovery process maps out existing processes, identifies gaps and then models a future state. It’s quite a formal process, and it’s chargeable. It’s the best way of getting the necessary details, and the banks appreciate the equal robustness of our engagement.
TFM: What potential does AI have in unstructured areas, such as invoice processing, and how can it spill into other areas?
FT: AI, by itself, isn’t really fool proof, just like humans. But, with a combination of human and artificial intelligence, we can get much better accuracy in the capture and verification stages of data, and that ensures that you’ll get higher automation levels. Today we do that just for invoice processing, but we’re going to extend it to a whole raft of other back-office services.
Automation has basically pretty well reached its limit with structured data, so anything structured can easily be automated. But you’ve got this next low-hanging fruit, which is the unstructured data. And the technology is there as long as you provide some context to it. So, if you say ‘understand all the variables of this room and the context’, it’s very difficult for AI to make sense of that. But if you say, for example ‘you’re only going to receive an invoice’ and then give the learning around all the different variables associated with that invoice, in the next step, it can take another form, say, a mortgage application, and start comparing documents.
The combination of AI and human intelligence in the handling of unstructured data started creating amazing synergies. The services available today are all based on the technology that was available in the past. The new technologies and approaches, new combinations of artificial and human intelligence, are going to unlock new services that weren’t previously possible. And that’s where our strength lies. Our core strength is our people, who do the end-to-end consulting, implementation, support and integration. The software, by itself, does nothing, until it’s integrated with the people, processes and the technology. The key area we often focus on with this platform that has open APIs, is integrating with the customer’s systems, and evolving them. That’s the key challenge, and that’s how you get the best results.