Raised on the warm, white-sand beaches of Costa Rica, Paz Gonzalez now resides near the snowy white mountains of Vancouver, British Columbia. As an outdoor enthusiast, Paz enjoys the Canadian weather through activities like snowshoeing, skiing, and hiking.
In this Q&A, Paz explains the difference between Base Automation and Touchless Automation and how machine learning (ML) works at Conexiom.
Describe your day as a software developer.
My typical day focuses on machine learning and Base Automation. My team and I have daily meetings to discuss what we accomplished the day before, what challenges we encountered, and how to overcome those difficulties.
If I’m working at the beginning of a release, I focus on implementing new features for the product. At the end of the release, I focus on perfecting the product so it’s ready for production.
Conexiom’s automation platform contains two product offerings: Touchless Automation and Base Automation. How do they differ?
Here’s how the Touchless Automation solution works: When a trading partner sends an order, Conexiom performs a data extraction on the order. Conexiom processes more than 80% of incoming orders without human intervention, which is why it’s called “touchless.”
After it processes the order with no exceptions, Conexiom creates an output file and submits the order via Secure File Transfer Protocol (SFTP) or another delivery protocol. Finally, the seller processes the order, and the buyer receives their order. This process works best for high volume customers.
The Base Automation solution, on the other hand, is a simpler process. It is a complementary product offered in conjunction with Touchless Automation, and it accommodates low volume customers. Base Automation, like Touchless Automation, receives the trading partner’s order via email and performs a data extraction.
At this point, an alert is generated in the system for a customer service representative (CSR) to complete the data extraction within the MyEditor application. Once the CSR reviews the document, they can update any information before submitting the order. The CSR can also decide whether they want to move the trading partner to Touchless Automation based on order frequency.
Base Automation is a valuable solution because 75% of documents that are assigned to it are successfully configured. Therefore, Base Automation users can quickly fix and approve documents to submit orders.
Base Automation is also faster on orders with more than 10 lines. For CSRs who perform manual order entry, Base Automation saves time on multi-line orders by reducing the amount of manual entry and errors.
How does ML work at Conexiom?
The first thing we do to teach the ML is collect data. We filter that data into an intelligent algorithm to teach the ML the steps we want it to take. Data fields are identified by three elements: labels, positional values, and value formats.
The ML finds the right label and looks for its value, usually located below or to the right of the label. It verifies the value and matches it to the expected format (words, numbers, single or multiple lines, etc.).
We code that data into the ML library and test it. After testing, we verify the information that should be recognized as positive and negative results.
For example, with a purchase order (PO) number, the ML looks for labels equivalent to the PO number. Labels might say “P.O. #” or “Purchase Order Number.” The ML recognizes those labels as positive and other labels like “Quantity” as negative.
Then, we use a business logic and lookup layer based on account logic and previous iterations of documents to update and validate the information on the order. This makes it so the CSR doesn’t have to make the same changes over and over.
If you could automate anything to assist you in your personal life, what would you automate?
It would be amazing to automate cooking and cleaning. I recently re-read the entire Harry Potter series, and I wish that I could wave a magic wand to chop potatoes while doing other things. That would be perfect.
Hear More Conexiom Voices: