![\"1800](\"https:\/\/intercom.com\/blog\/wp-content\/uploads\/2017\/02\/1800flowers-394x700.png\")
<\/p>\nWhen consumer internet arrived in the 1990s, it ushered in a new age of communication.<\/p>\n
Free internet chat applications like ICQ, MSN and AIM were limitless and meant we no longer had to rely on expensive and metered phone calls. Yet it was text, not voice that was to dominate in the years to come. Slow dial-up internet meant we ended up typing to each other, not talking to each other.<\/p>\n
So in a world of 4G and fibre broadband, why are we still clinging to text interfaces in unsuitable contexts? Nowhere is this clearer than with chatbots. Text-based messaging between humans is useful because it\u2019s asynchronous \u2013 we\u2019re happy for the conversation to carry on later. But chatbots are different. They represent the return of the command line interface \u2013 call and response \u2013 so we expect communication in real-time.<\/p>\n
2016 saw thousands of new chatbots emerge, and while texting with them might have seemed fun, they consistently delivered sub-optimal value. Typing on small screens with friends is convenient, but text-based interactions with chatbots less so. The effort rarely reaps the rewards we expect:<\/p>\n
<\/p>\n
Is Facebook’s 1-800-Flowers bot really making things easier?<\/em><\/span><\/p>\n Text-based messaging works best with a human on both sides of the conversation<\/p>\n Imagine walking into Starbucks and ordering a \u201cVenti, half-whole milk, one quarter 1%, one quarter non-fat, extra hot, split quad shots, no foam latte, with whip, 2 packets of splenda, 1 sugar in the raw, a touch of vanilla syrup and 3 short sprinkles of cinnamon.\u201d<\/p>\n Now imagine again if Starbucks insisted each part of the order was written by customers on a separate Post-it note? If Starbucks had never employed humans all over the world parsing each and every voice-based order, they would not be the global success they are today.<\/p>\n Therein lies the challenge with today\u2019s chatbots. The way we ask for their help \u2013 through text, not speech \u2013 is unnatural for humans. Which makes sense. We\u2019ve been muttering and grunting at each other for at least 100,000 years, but writing things down for only 5,000 of them.<\/p>\n Text-based messaging works best with a human on both sides of the conversation. Humans understand the context, intent and sentiment of other humans. A conversation with a text-based chatbot lacks human emotion, clarity, and urgency. Surely there must be a better way? Enter voice chatbots.<\/p>\n To explore ways in which voice can make chatbots more useful, let\u2019s use a simple example of a normal everyday conversation between two humans.<\/p>\n It\u2019s Friday afternoon, and Jim calls Jane. Both work on opposite sides of New York City. They chat for a few minutes to plan grabbing dinner and drinks, and maybe take in a Broadway play after work.<\/p>\n Jim adds Concierge-Bot to the conversation with a simple command<\/em>:<\/p>\n Jim<\/strong>: \u201cHey, Concierge-Bot\u201d<\/p>\n Concierge-Bot<\/strong>: \u201cHow can I help?\u201d<\/p>\n Jim<\/strong>: \u201cJane and I would like to meet near West 47th Street and 9th Avenue for drinks at 5pm, have dinner close by at 6pm, and would love to see a comedy show at around 8pm.\u201d<\/p>\n Concierge-Bot<\/strong>: \u201cBar Centrale serve cocktails, and Obao serve an Asian fusion menu with availability at 6pm for 2 people. Drunk Shakespeare have availability for their 8pm show. Would you like me to reserve these options for you?\u201d<\/p>\nVoice bots drawing inspiration from offline conversations<\/h2>\n
![\"Google](\"https:\/\/intercom.com\/blog\/wp-content\/uploads\/2017\/02\/Google-Assistant1.png\")
<\/p>\n![\"Cortana,](\"https:\/\/intercom.com\/blog\/wp-content\/uploads\/2017\/02\/cortana.png\")
<\/p>\n