There has been a lot of discussion and curiosity in the Accounts Payable world about the role of Artificial Intelligence, or “AI”. In fact, it was one of the hottest topics at IOFM’s annual APP2P conference last month. While AI may have taken a more prominent place in the decision-making process for AP Automation, it is nothing new in the software world.
Other articles you might find helpful:
“Artificial Intelligence” is broadly defined as “the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.” By that definition, even calculators qualify as Artificial Intelligence.
Here at DataServ we have been leveraging AI before it was a commonly recognized acronym...certainly before anyone really believed in self-driving cars and self-aware robots. In fact, any SaaS automation is, at its core, artificial intelligence, so it’s one of the top tools for creating value in our company.
We’ve seen some amazing developments in AI, some of which have been game-changing for SaaS automation. And in other ways, a fatality involving Uber’s self-driving car demands caution on the approach to artificial intelligence; it’s a life-saver for routine tasks, but it must recognize exceptions to the norm and leave those to the human in the driver’s seat.
Assessing and integrating AI technologies is one of the primary responsibilities our Director of Technology, Drew Gillow, is charged with.
Drew spent over 20 years at Monsanto, including several years in logistics optimization, before coming to DataServ, and of course logistics optimization relies heavily on AI. As Drew says, it’s possible that a human could do a better job than the computer if they spent the time and applied all their judgment to the job. But in logistics they don’t have the luxury of time, so the computer’s job is to find a good solution and to find it fast. That leaves people free to focus on jobs requiring the best possible solution, which means even better solutions for the jobs that matter most.
That is exactly how we believe AI should be used in the automation of financial processes as well. Some tasks need to be optimized for the best possible solution. Some tasks need to be maximized for the best fast solution. A computer can’t make good decisions about everything, but the more decisions we can safely delegate to AI, the more energy and focus people can give to things that have to be perfect.
Of course, you have to be the one to decide what can be optimized and what must be perfect. That’s why we not only leave your people in the driver’s seat when it comes to managing AP exceptions, we also allow you to define your tolerances and triggers so that the software knows what should be flagged as an exception. Granted, a car with no human driver is very different than AP Automation. But there are some parallels that are instructive when you’re comparing AP Automation partners.
Because philosophy drives process, before you ask about any prospective partner’s technical expertise with AI, it’s important to understand their philosophy about delegating decisions to machines.
For instance, there are two basic principles that form the foundation of the DataServ philosophy and approach to using AI. One is “don’t get ahead of the technology.” Just because the machine can be taught to do it doesn’t mean the technology is dependable enough (yet) to put the machine in charge of the decision. So we put a lot of our focus on “baking in” best practices. During Initialization, we work with you to set up checks and balances, define exceptions and tolerances, and document process and workflow. That means that the tasks that are part of your automation have been vetted by humans before they’re delegated to even the most advanced machine.
The second is “delegate, but don’t abdicate.” Self-driving cars do have a safety driver, a human whose job it is to take over if technology fails. If that human doesn’t intervene when necessary, that’s abdication, not delegation.
Our philosophy is that both your team and the AI need to have crystal clear instruction on when to stop the process until a human takes over. That means that you must have control over approved tolerances and discrepancies that trigger an exception flag. The software may make recommendations, but when an exception is detected, a human is required to review and approve before the process moves forward. By baking in best practices, we can prevent “accidents” due to machine error and greatly reduce accidents due to human error as well. That is delegation at its best.
Which brings us to oversight, or the ease with which you can review the decisions made by the software and the decisions that are pending human judgment. How efficiently and effectively does your computer interface with the human “driver?” Just as the dashboard of a car is designed to allow the driver to assess all systems at a glance, the dashboard of your AP automation should allow your team to quickly assess the status of work in progress and urgent items needing their attention.
With that in mind we’ve designed our dashboard to be even more intuitive and informative. Users can set their own thresholds and triggers and the dashboard highlights areas that need attention according to individual priorities. This helps users structure their time appropriately and provides managers instant and easy access to the information they need to balance workloads and accountability. And because it is a SaaS platform with mobile functionality, users and managers can perform that oversight no matter where in the world they might be.
Ultimately the success of any automation project comes down to the trust you and your teams can put in the technology. Which we find has less to do with the dependability of the AI than it does with how well the people involved understand the decisions AI is making. That is why we put so much emphasis on training and change management. It is during that phase of your Initialization that trust is cemented between your team, our team, and the automated process that is AI.
When people know, understand, and have input into how the software is designed to make decisions, when they’re comfortable with what the software did or recommended, and when they have ready-access to the data and details involved, then they are more likely to work synergistically with AI. If they don’t have that information and that level of trust, then they won’t feel confident letting go of those tasks that can be handled by the software which defeats the purpose of automation.
Our goal with AI is not to put software in charge of all your decisions. Ultimately our goal is to minimize stress and maximize productivity by making it easier for you to adhere to best practices, reduce human error, have more real time access to your data, and allow your valuable people to have the freedom to focus on the decisions only they can make. That means our system is designed to make driving and navigating a whole lot easier, while still leaving you firmly in the driver’s seat.