Ever since Franklin D. Roosevelt’s time, there has been a lot of emphasis on what a US president can accomplish in their first 100 days in office.
Although choosing the SEMINE platform might not be comparable to taking over the White House, we think it is important to get out of the starting blocks quickly and make the most of the enthusiasm that finance department employees tend to feel in the initial period after the system is up and running. It also shows the rest of the organisation the benefits and effects the new accounts payable (AP) automation system can bring.
Despite what many people think, getting started with SEMINE is not some complex IT project. We start with a kick-off meeting, make some technical preparations and set up the various integrations. Then we run a day of setup and training for superusers, and you’re good to go. If your company uses one of the most common cloud-based financial systems with relatively standardised integrations, you may to be able to go live less than a month after signing the agreement.
To achieve good results quickly, there are some preparatory steps you can take. Since SEMINE uses Artificial Intelligence (AI) and Machine Learning, there are some trade-offs that need to be made: Which tasks should be prioritised first in order to free up as much time as possible and as quickly as possible? Should the algorithm be trained based on how invoices have been posted in the past, or is it better to start with a clean slate to ensure that you don’t teach the AI engine using previous errors. Should strict automation rules be set for how the algorithm learns, or should it have as much freedom as possible?
Several actions can be taken even before the system is installed and operational. For example, if you make an effort before the project starts to get suppliers to send well-specified invoices via e-invoicing or establish standardized processes for invoice processing, you will have a much shorter path to achieving a high level of automation. Such measures provide the AI with more reference points to work with and make the path to a successful result shorter.
READ MORE: Fast adoption with a structured and efficient implementation process
One thing that surprises some people is that in the initial phase after starting, it may be best to let the AI work on its own without too much rule-setting. The AI algorithm learns through repetition, and it needs to experiment a bit before improving based on more concrete feedback. This varies from company to company – in other cases, it might be more optimal to give the algorithm more defined guidelines for training the AI engine – but this is something we determine together during the onboarding phase.
The goal of any SEMINE project is always to achieve as full automation as possible. Usually, the degree of automation reaches 70–90 per cent fairly quickly but can then become a little more demanding.
Each step frees up time to take the next step: Which invoices from which suppliers does the algorithm always get right, and can the authorisation workflow for those invoices be automated? Which suppliers should be asked to provide greater detail in their invoices? What internal processes should be adjusted to better facilitate automation?
Throughout this work, SEMINE closely observes what works and what should be improved. The details are presented in a simple overview that makes it easy to see what the next step should be and what measures should be prioritised at any given time – depending on the effect it will have or how easy or difficult it will be to address. This overview tool also has functionality to simplify all necessary tasks, making the transition to hyperautomation as easy and painless as possible.