Why are some bots great and others meh?

Not only did the chatbot took in my personal details, but also provided a plan for the weeks to come, connected me to a real personal trainer for accountability and provided notifications and even motivations that are personalized to me.

I can’t help but wonder, “why hasn’t this become the norm for the fitness business reaching out to weekend bums like me?”

All bots automate but not all are smart

The fitness chatbot was one example of how advance bots can be today. Bots today can do more than provide information but recommend and transact as well increasingly in business functions anywhere in recent years. I guess that could be the reason Google records a three times increase in search on the term ‘chatbot’ globally in the last three years.

In my work in discovering the world of chatbots and virtual assistants technology and speaking with partners and clients, I realize that while organizations have ambitions, wanting bots to do more, they face a dilemma of whether they could?

If you are like me who love fiddling and testing chatbots, you will realize that while most bots function ok, the ones that truly stand out are the ones that can deal with a wide variety of scenarios, and even relate to me, so much so that its as it’s a human on the other side. So why are some bots better than others?

The good ones are made for the multi-input world

If we take a look at how some service sectors that deal with a wide spectrum of customers, for example, government agencies you’ll find that they probably understood that being communication centric and language-centric isn’t the same thing. And the former is what they did to serve a larger group of audiences. Being communication centric means focusing more on the users’ preference to relate.

Features within bots like voice to text, allowing an elderly perhaps to speak and be understood by a bot instead of having to text. Or being able to take a picture and upload, and the bot could scan, identify, match and confirm it is, for example, a payslip or a valid signature.

Even today, there is a preference to create pluggable bots that can work on customers’ native social channels like Whatsapp, WeChat or Line.

The good ones speak like your best friend

This is necessary for South East Asia where within a country, ethnic slangs and demographic nuances exist and sets a high bar for any bot to be deemed human-like. Features that allow bots to learn and remember common nuances like acronyms. Features that recognize truncated statements or short forms as well are particularly useful for high volume customer service counters.

Today bots that can self-learn and self-remember to a point they could recall what you said a day ago and point you in the right direction like your best friend, is becoming the expectation for fast moving, labour intensive organizations like fast food chains and retail that are looking for a robust and reliable service solution that perform at speed and scale at a fraction of their current setups.

The good ones talk to each other

Think about what’s happening today when you call a bank to ask about a loan. You are passed across departments, deal with long pauses, till the right service officer with the right actions finished with you. It’s worse when mistakes are made or sub-par recommendations are given. This process repeats itself until a customer is satisfied or worse terminates with the bank. Today’s matrix organizations like banks where a customer is a customer across multiple departments - mortgage, car, insurance..etc. The expectation of customers is demanding organizations to be more connected than ever before. Banks that are planning ahead are orchestrating multiple bots linked across data systems to ‘speak’ with one another towards a customer-centric vision where a single bot interface can confidently deal with customers at a fractional of the time and at the speed, security and precision demanded.

A good bot is like the infinity gauntlet in the Avengers. Start collecting the stones.

Not all bots are built equal. It’s tempting to take an off the shelf bot solution for brands today who move fast. But with the rising business complexity and customer expectations, failure is costly both financially and in brand equity. As I reflect, I realize a good bot with real-world ready intelligence is like the infinity gauntlet with all its stones (this one’s for the Avengers’ fans). Each stone represents a different artificial intelligence discipline and architecture needed to make it powerful. All bots process information but not all are smart enough for the real world. It’s all in the build.

Joshua Kwah is the marketing director of Taiger, an artificial intelligence company that uses human-like logic to read, understand and extract information. It is headquartered in Singapore with offices in Hong Kong, New York, Madrid and Mexico City.