With COVID-19 disrupting many supply chains across the country, companies are doing everything in their power to strengthen the relationship they have with suppliers. AI-enabled payments ensure organizations pay their suppliers on time. When supply is lo, vendors will prioritize companies that are easy to work with and pay them promptly.
To learn more about the ways AI-enabled payments could benefit organizations facing shortages within their supply chain, we spoke with Lloyd Humphreys, the Principal Product Manager for Data & Analytics at Tradeshift. Tradeshift offers supply chain management and automated invoicing solutions to improve supply chain-related processes.
Jenn Fulmer: What are the benefits of using AI-enabled payments?
Currently, accounts payable (AP) teams spend around a quarter of their time chasing and correcting invoice exemptions, and more than four in five invoices require some degree of manual intervention. That explains why it costs an average of $11 and takes eight days to process every invoice.
We’ve seen how a combination of digitalization, supported by AI and automation can help reduce the cost of processing invoices by a factor of five and slash the average processing time to under three days. Even then, that’s only scratching the surface of what AI-enabled payments can achieve.
Which parts of the payment process can AI automate?
AI can automate potentially any part of the payment process, but that doesn’t mean businesses have to (or should). Invoice processing is very sensitive and fault-intolerant, so AI needs to be thoroughly trained and tested before you relinquish control to the technology.
At Tradeshift, we recommend that businesses start small while you teach the AI how to recognize common errors. A perfect “robotic” task for AI is performing line matching between invoice and purchasing order or coding. This is mundane, repetitive, and time-consuming work for people, which goes a long way to explain the high rates of human error: one recent report estimated that 23% of all invoices contain a mistake requiring manual review and reprocessing.
In theory, AI can automate the entire payments process, although in practice businesses will want to keep at least some control. The problem is that almost every AI for accounts payable is a case of “on or off”, but that’s changing. At Tradeshift, we’re the only provider to enable businesses to finely dial up or dial down the degree of automation depending on what they’re comfortable with. So you can decide whether the AI operates conservatively — always querying and flagging items it’s unsure of if you simply want to provide people with a digital assistant — or give it relatively free rein if you’re going guns blazing for automation. The great thing is that this means you don’t have to choose between the two poles, but instead choose the balance that’s best for your needs.
Can automated payments offer a competitive advantage for businesses working with suppliers during a shortage?
Absolutely. If automated payments only offer benefits in one direction, it’s not being done right. When an organization is in control of AI, it can do away with the approvals process entirely on certain recurring or low-value transactions from a trusted vendor.
Too often, payments are a roadblock, a chokepoint. By automating elements (or even the entirety) of invoice processing, businesses will remove backlogs and pay their suppliers faster and with fewer errors. But that’s only the beginning of the benefits payments automation can bring. It enables businesses to forge deeper, more trusting relationships with partners — for example, by extending lines of credit or finding other ways to reward partners who pay early. Automating payments is the first, critical step towards building this foundation of trust; what comes afterward in terms of deepening and extending these relationships, is very much up to businesses’ imaginations.
Also read: Top 5 Benefits of AI in Banking and Finance
How much coding knowledge do companies need on their team to implement AI-enabled payments?
If you need to get involved in the coding, something has gone very wrong with your AI payments application. The future of AI is about enabling anyone in any job role to start doing their job better without having to retrain or spend valuable time getting elbows deep into coding or other technical aspects.
In everyday applications, AI should stand for Automated & Invisible. In invoicing, one of the key benefits should be straight-through processing, which means that AI is mapping documents and finding the associated purchase order automatically.
That’s not to say that businesses don’t need to invest in training their employees on the new technology. But this is about process, not about technical skills. Workers need to understand that AI is only as good as it’s trained to be and that it’s never a case of switching it on and letting it get to work. The teaching and training stage for AI is absolutely critical to delivering AI’s full potential and avoiding costly errors. Organizations need to ensure their employees are absolutely clear on how the entire process works, including the parameters for full automation.
Once that is in place, human operatives should only have one job: to set the different tolerances for how AI operates, between full automation, conservative operation, or anywhere in between.
What kinds of security protocols are in place to keep transactions safe and private?
Our own policies include third-party audits that enable us to provide our customers reports validating the security of the platform with standards such as SOC 1 Type II, SOC 2 Type II, ISAE 3402 Type II, Payment Card Industry (PCI-DSS) Level 1, and ISO 27001.
We’re also certified under the EU-US and Swiss-US Privacy Shield program which covers cross-border data transfers to the US and was similarly certified under the preceding program, the US-EU Safe Harbor.
Standards are important, but they don’t cover everything. For example, there’s a potential security loophole when a provider uses template machine learning (ML) models, which may retain sensitive data from other customers. That’s why all our ML models are trained entirely from scratch for each individual company.
We understand that businesses may be nervous about the security implications of adopting new AI payment technologies. But the real danger is doing nothing in the face of fraud and other cybercrime. Given that humans are usually the weakest link in any security posture, automating processes can remove important opportunities for attackers to exploit. There are other ways technology can play an important role in the fight against fraud. For example, at Tradeshift we’ve developed the My SiS-id app, a community-driven, blockchain-based platform that authenticates and secures banking payment contact details to create a tamper-proof, shared ledger of verified bank data.
Read next: How to Prevent Third-Party Vulnerabilities