Some of our more tech-savvy peers may remember the first generation of chatbots, such as ELIZA and ALICE (Artificial Linguistic Internet Computer Entity). ELIZA was born in the mid-1960s and ALICE, her 21st century upgrade, some 25 years ago. Thankfully, chatbot technology has since evolved using Natural Language Processing (NLP) techniques to create more useful smart bots that can mimic human characteristics and thinking. With voice search rapidly increasing in popularity, how do we continue evolving our chatbots to ensure that our customers have an increasingly comforting, ‘human’ experience? The good news is that you’re in control of what you build from scratch.

The success of any smart bot depends significantly on the persona that has been developed for it. This is why you have to ask yourself the hard questions upfront. ‘What would my bot look like if it were human: hairstyle, eye colour, what would it be wearing, and even where does it live? If your bot is resident in New York, for example, you could consider including a few relevant facts about the city in your response bank. Your target market will probably dictate whether it should be male or female, or you could opt for a gender fluid or gender neutral bot. What level of emotional empathy should it have? Empathetic virtual agents such as Ellie have confirmed that some of us prefer to speak to a bot instead of other humans. This is because certain people feel safer in the knowledge that they are not being judged and that their personal information will not be leaked.

What kind of language should it speak? If, for example, your bot is helping people to invest in fine art, their speech should be infused with phraseology that high-end art dealers may use in describing the style of selected artists. If you happen to be the CIO, it may be a great idea to consult with those arty colleagues on the first floor.

And sentence structure is an important consideration. Should your bot be formal and grammatically correct at all times (‘Yes, I understand’), or use friendlier language constructs (‘Okay, I get it’). Some bots now feature emotion detectors which read facial expressions and recognise feelings such as anger, frustration, joy and happiness.

Once the author has thought through the nature of their ‘new best friend’s’ personality – its look, style, feel, tone and manner – a detailed document should be circulated to all response-bank and tech developers who may be involved. You will meet with much more success if you incorporate their considerations and suggestions.

Unlike keyword recognition bots, smart bots ‘grow’ over time. This enables them to suggest your favourite hamburger topping (cheese no pickle) and that side of onion rings, and have a record of your delivery address and payment details. US Research has shown that 80% of customers are likely to use a product or service again if the service was exemplary.

Well, with bots such as Endurance, which is designed to chat with patients who have dementia, and CASPER, who engages in midnightly conversations with insomniacs, bots are here to stay. Our job is to inject as much humankind as we can.

Hey Siri…Alexa…Bixby? Can you hear me?


Need help with communicating to your customers? Contact 

Altron Bytes Managed Solutions

1 Vlak Street, Selby Ext. 5, 2092, Gauteng,
South Africa
PO Box 3591, Johannesburg, 2000,
Gauteng, South Africa

+27 (11) 373 4000

View our range of COVID-19 Solutions

You have a vision.
We can deliver the innovation
to take you there.

1 Step 1
Full Name
Contact Number
Business Name