We're aiming to have our product progression documented to help our customers easily see the great things we’ve accomplished, and we also want future customers to know that we’re not merely innovating in AI, we’re also adding disciplined product enhancements every day that deliver the best and most advanced customer experience possible.
With that, the DigitalGenius September 2018 release notes:
Language Filtering for Zendesk Macros
Managing a multilingual customer service team is difficult enough; you need to have a group of skilled and talented professionals, who need to understand the brand tone of voice in each of their languages. It’s critical they have up-to-date, consistent content at their fingertips.
That’s where our Language Detection and Filtering technology comes in. Prior to this, agents would be presented with AI-suggested macros in a mix of macros, dependent upon the AI score. DigitalGenius is now able to structure macros in separate languages, making it easier and quicker for agents to use the suggestions or search through them in the relevant language of the customer.
Many of our customers have multilingual macro libraries, as well as agents who serve customers in more than one language. This requires quick access to separate libraries. We’ve released a dropdown at the top of the macro recommendation window in which agents can now toggle between languages, and select the one they require. (See the image above).
Detecting the correct language and presenting it in a streamlined manner to the agents is not only a great time-saver, it also significantly improves the quality and accuracy of the AI predictions - making the your AI model smarter and quicker.
APIs for Confidence Thresholds
With any new customer service technology there's always a tug-of-war between improving efficiency and the quality of the customer experience. It can be even more pronounced with artificial intelligence and machine learning on board. Customers regularly ask us, How can we automate substantial customer service volumes, yet still ensure our customers receive the great experience we're known for?
Since our inception we’ve handled this with care, and have designed our algorithm with high pre-set confidence thresholds to ensure AI coverage exclusively captures highly repetitive and straightforward tasks - significantly reducing the risk of a customer getting an incorrect or irrelevant answer.
However, that approach is not appropriate for all customers at all times. With that in mind we’ve released APIs that allow customers to lower the confidence threshold of the AI, thereby increasing the coverage of what the algorithm will automate. This could be appropriate for high volume scenarios, or for certain peak times of the year; for example, when it might be better to provide a response even more rapidly than normal.
Our customers will always know their organizations , their priorities and their appetites for risk better than we ever will. Our job is to empower them to achieve success, in whatever guise it comes. We’re delighted to now start offering this customizable functionality in our September release.