reps assist customers who contact the company via text message, the web app, social media, or any other text-based space.Expert Assist gathers information, categorizes the message, and serves up relevant articles from T-Mobile’s internal wiki to help the customer service rep resolve the customer’s issue—all before the rep even receives the initial message.
Nolis’ team built a robust model that “listens” to conversations for signals that a person wants to leave the AI experience and reach a human. For example, if someone uses the words “help,” “real person,” “operator,” or similar language, they’re presented with an opportunity to talk with a human.With conversational AI like a chatbot, make sure there’s one tone of voice, and that it’s the voice you want representing your brand.
“Normally when you’re developing software, faster is better,” she explains. “But when you’re developing conversational AI, even if you were able to generate a three-paragraph response within the enterprise standard of under 250 milliseconds, you need to build in a pause before the answer. Otherwise it’s a jarring experience. So you build in that pause for the comfort of the human.”Nolis points out that it takes tremendous agility to account for the unexpected with conversational AI.
In addition to maintaining high standards, using human metrics offers another benefit: highlighting issues that need to be addressed. “By measuring AI this way it can show us where to improve. For example, if our bots aren’t empathetic enough to customers who are affected by the pandemic, then we know we need to do some language updates to apply a little more human intelligence to the chatbot.
TMobileBusiness
ประเทศไทย ข่าวล่าสุด, ประเทศไทย หัวข้อข่าว
Similar News:คุณยังสามารถอ่านข่าวที่คล้ายกันนี้ซึ่งเรารวบรวมจากแหล่งข่าวอื่น ๆ ได้
แหล่ง: Forbes - 🏆 394. / 53 อ่านเพิ่มเติม »