Connection Uses Fisent BizAI to Tap the Power of GenAI Models for Process Automation
How a Fortune 1000 global IT solutions provider dramatically reduced the time needed to process customer orders and improved the experience for its customers and sales teams with Fisent’s Applied GenAI Process Automation solution, BizAI.
The Challenges
Before adopting Fisent BizAI in 2024, one of Connection’s “good challenges” was tied to the sheer volume of purchase orders (POs) it must manage (hundreds of thousands of them across tens of thousands of customers). Don’t pity Connection’s industry-leading market position, but instead appreciate the high-pressure role of the accounting and credit teams have in upholding Connection’s commitment to customers: expeditiously reviewing 50 to 60 thousand incoming sales orders against the POs on file to ensure that Connection is providing goods exactly as specified, at the agreed price, to approved buyers, all while meeting or exceeding promised service level agreements.
The approach that Connection was using to match these documents relied heavily on human review. Although they also explored traditional automation approaches involving document mapping, even a tiny issue could cause a mismatch, such as if an address isn’t input using identical characters. When issues arose, it might take hours (or days) to sort out, delaying shipments to valuable customers anxious to get their orders.
“The timing just didn’t work at scale, especially when we have such a diversity of customer documents we need to understand and analyze,” explains Jason Burns, Connection’s Senior Director of Process Optimization and Transformation. “With traditional approaches, you’re often trading developer or configuration time in exchange for the gains you could get by fully automating the match process. This is why we looked to apply GenAI for process automation.”
What Connection needed was an automated, standardized, and highly secure solution that would enable accounting teams to manage by exception. GenAI held the promise of efficiently scrutinizing huge amounts of data and offloading much of the human burden. The search for a vendor, however, proved frustrating at first, because those with whom Connection met — even the larger players in the order-processing space — lacked the necessary AI technology. “I was sort of surprised that they didn’t already have these AI capabilities baked into their products,” says Burns.
The Fisent BizAI Solution
Connection knew it had found a promising partner and technology when it spoke to Fisent about its Applied GenAI Process Automation solution, BizAI. Once Connection explained the use case and the data corpus to be executed, Fisent took only a few days to present a results-oriented proof of concept that Connection could demonstrate to business stakeholders.
“My reaction was disbelief,” says Burns, remembering Fisent’s presentation. He was especially amazed by how quickly BizAI could be implemented and how well it would solve Connection’s challenge. “I distinctly remember thinking to myself, ‘This is transformative in its potential. We now have the ability to help our customers in a way that is a real market differentiator.’”
This was Connection’s first venture into implementing an AI solution, so Burns anticipated having to calm people’s fears and insecurities about changing their routines, trusting a new technology, and taking on risk. But convincing account managers, accounting and collections teams, and IT was a relatively easy lift, because Fisent had tightly scoped the project, the cost was reasonable, and the upside for order management and customer satisfaction was so clear.
Burns adds, “Fisent did a phenomenal job walking us through all the risk analyses and helped us know that we were doing the responsible thing instead of just the exciting thing. Frankly, they made it so easy for us to engage with them, it was almost a no-brainer. They were extraordinarily engaged in making this successful and took what felt to me a very personal interest in delivering success.”
After Connection received feedback from stakeholders, it chose to develop integration components with its ERP system internally, although “Fisent really carried close to, if not all of, the balance of the work, so there wasn’t much for us to implement,” reports Burns.
Then Fisent produced what Burns calls “the magic piece.” It configured the application to provide line-by-line analysis of every piece of data being compared, enabling matching that was nearly identical to how humans would complete the task, but with lightning speed.
Rigorous testing with the accounting and IT teams followed. Connection collected data throughout the process to determine how the solution was performing in real-world scenarios and to eliminate any difference that existed between the test environment and actual applications. Based on assessments of the solution’s recommendations output, Connection made some small tweaks, “and we were in full production about 90 days after we kicked off the project,” explains Burns.
Business Outcomes
Matching accuracy
After adopting Fisent’s BizAI, the accounting team and others at Connection were surprised, even shocked, by the matching accuracy the solution had mastered. Instead of the 60 or 70 percent accuracy they expected, they were seeing matching accuracy approaching 100 percent.
“We were all impressed at just how good the AI performed right out of the gate,” says Burns, “and it feels like BizAI is finding things that humans may not normally catch.” For example, the Fisent team tuned BizAI to be very judicious before recommending that a sales order move on to the next step in its journey. In fact, the Connection team observed that when the BizAI solution found a minor discrepancy, it would more intensely scrutinize other properties as part of its assessment task.
Improved speed
BizAI was able to add efficiency to the system by greatly reducing the amount of time accounting teams spent comparing documents. When a PO and sales order did match, the order was progressed immediately and automatically. For PO-sales order mismatches, Connection saw a 98 percent reduction in processing time. What’s more, improvement was realized across all orders because Fisent BizAI can read a plethora of document formats, such as PDFs, Excel spreadsheets, Word documents, image files and even.msg email files and their related attachments, which are all common ways customers prepare POs and submit orders.
Security
Connection’s concerns about security and data privacy were solved, as well. Fisent’s technology is fully encrypted, protected, and stateless, which is of particular concern when employing large language models (LLM). So whether Connection’s data is in transit or at rest, “our data is protected. None of the information sticks on an LLM in a place that we might not intend it to. We went through many weeks of security review and found BizAI to be a thoroughly secure solution,” explains Burns.
Key Takeaways
Connection teams continue to benefit from BizAI’s automated processing of customer orders. Account managers receive real-time feedback when there’s an issue with an order. Credit teams who used to have to engage in guesswork can feel confident in Connection’s output. Collection teams no longer spend so much of their time chasing payments and now can focus on value-added, strategic decision-making.
“They absolutely love the technology,” exclaims Burns. “Instead of doing work that is mundane, they’re able to focus on their customers, deliver better service, and concentrate on things that are of much higher value.”
The improved efficiencies leveraged by BizAI delights Connection’s sales reps — who can get orders approved immediately or flagged when order remediation is required — and their customers — who receive orders sooner and appreciate what they perceive to be exemplary support from their account reps. What’s more, finance leaders are pleased that the high-performance solution can provide an exceptionally detailed reporting and audit trail.
Since adopting BizAI, Connection has been talking to other functions in its organization about how the solution could increase efficiency by automating their business processes. Within a few weeks of demonstrating the order matching use case to the broader enterprise, Burns has prioritized about a dozen additional use cases that will implement the BizAI solution.
Burns says the company is “extraordinarily happy” with the outcomes and adds that the improvements BizAI has brought to Connection’s order processing will empower the enterprise to be more competitive in the marketplace.
He appreciates, too, that Fisent has provided guidance and support throughout the entire process, from their first sit-down through implementation. “Fisent has gone above and beyond in terms of providing consultation and input. They’re partners, really, not just a vendor for us.”
