I want to take a day off work. I go to the internal Oracle employee site and click through to “Add Absence.” So far, my user experience has been wonderful and I’m looking forward to requesting a day off. Now I click the “Absence Type” list of values (LOV) and see the following options:
In over 20 years working at Oracle, nearly every time I’ve used this app has been to request vacation. I’ve never used eight of the eleven options, and “military service” and “marriage” are unlikely choices for me. But I still have to scroll to the bottom of the list every time. Is this a real problem? No, not really. I don’t take that many vacation days. It's still annoying though.
But what if you work in procurement and spend hours a day creating purchase orders that get shipped to Zurich? Scrolling down to "Z" in a static, alphabetized list 50 times a day can be vexing. Typing an exact match each time isn't much better. Or what if you work in tech support and create lots of service requests for Zoom? Or you’re responsible for filling out expense reports for the Zagreb office? It might only take a few extra seconds each time to scroll or type, but do it 50 times a day, five days a week, for 50 weeks a year, and it adds up. Multiply that by the number of people working at your company in procurement, tech support, or expenses, and suddenly it’s a big waste of time and money.
Redwood intelligent recommendations solve this problem one interaction at a time by suggesting input values based on the click history of the user and users like them in similar business contexts. For example, the new Smart Select Single component released in Redwood 23A displays suggested values at the top of the list of values highlighted by a sparkle bar and set apart from the rest of the list by a separator.
For me, every time I want to request a day off, “vacation” would be at the top of the list and highlighted. Two clicks, no typing, no scrolling, low cognitive load. “Military service” and “marriage” still appear down lower on the list because they aren’t as likely, but they don’t disappear because they’re not impossible.
Intelligent recommendations from the Smart Select Single component are provided by the Oracle AI Apps Suggested Input Prediction Services, which generate machine learning models based on application context, data entered, and the user’s job role, business unit, department, and other work information. Oracle doesn’t ship a static one-size-fits-all machine learning model and hope it works for each customer. Instead, the Prediction Services code factories build multiple models for every Smart Select Single component based on each company’s unique data footprint. For each customer, for every component, multiple competing models are built and automatically tuned multiple times each day, with the best model for each customer and component winning out.
Data scientists optimize the factories, not the models. But the algorithms aren’t entirely generic; for example, for businesses that use Oracle Fusion Cloud HCM, the algorithm creates a social graph from the org chart to infer most likely options for contact names, assignees, approvers, delegates, and invitees.
Want some other examples? Oracle HR Help Desk will use the Smart Select Single to suggest service request categories based on the title and description information entered for the service request. Agents and self-service users can both enjoy a highly personalized and productive experience and spend more time solving problems and less time entering data.
Want to see intelligent recommendations in your favorite app, or want to discuss how your favorite app might benefit from using the Smart Select Single component? Start a conversation with us on the new AI Apps page on Customer Connect. For more information on Oracle AI apps, visit AI Apps Embedded in Oracle Cloud Applications.