Long, long ago (technologically speaking), Sci-Fi writers and film-makers were predicting that humans would be able to talk to their machines and that these futuristic, curiously American-accented machines would respond in a curiously American monotone and react to instructions, making life sweet and easy for humans. Remember Hal from the 1968 classic, '2001: A space Odyssey', for example? Ok, bad example, but you get the idea.

Nearly 40 years on, and very few people, bar a few eccentric propellerheads, have had the 'pleasure' of talking to their refrigerators and microwaves, or telephoning their home PC from the rush hour traffic jam to instruct the heating to kick in ahead of their arrival home. Indeed, despite all the advances in various areas of technology, we have generally retained our quaint, Victorian Age QWERTY keyboards, since alternative input technologies have not proved reliable enough: if you have ever tried off-the-shelf speech recognition software to dictate a letter, you will know what it is to hate technology.

In the realms of consumer-supplier interaction, a vocal exchange with automated systems is almost as rare as hen's teeth. Unless you count screaming down the phone at musak or cursing the slowness of an option listing.

The public is slowly being introduced to automatic speech recognition to conduct basic transactions, but such encounters are generally fairly rudimentary. We are still being forced down strictly fenced-off pathways, negotiated using tersely-barked orders, or the regimen of menu-structured telephone number-pad multiple choices. And it ticks us off. What people are after is flexibility: to perform multiple tasks without having to re-navigate a menu system, or – better yet – not to have to navigate a menu system at all.

Then, there is the bugbear of repeating information. Repeating details previously punched into the telephone keyboard and regurgitating your situation to various agents was recently identified by Accenture as the second most annoying aspect of dealing with a Contact Centre (the most annoying aspect was being kept on hold too long). To say nothing of repeating information. See? Irritating, isn't it?

Improved speech recognition engines, advanced business rules technology and process-based software can address both of these weaknesses. Capturing information, recognising trigger words and phrases and storing key information delivered at point of entry for later use is now a reality. Human natural language is a fiercely complicated field (think not only international languages, but regional accent variation and slang), but the flood gates are now well-and-truly open, and we all know what happens with technology once the ball starts to roll.

The emerging technologies are not only better able to understand everyday language, allowing for a natural, human-like dialogue between consumers and computers, they also recognise nuggets of information which can help determine—in advance—what the customer will want to do next, and store information for later use. Thus, repetition is effectively eliminated.

Already, automated dialogue systems are being developed and implemented for the most visionary organisations. These systems are able to understand instructions delivered using natural speech and/or the electronic written word. Considering the benefits and the public's increasing impatience and dissatisfaction with 'traditional' call centre approaches, the adoption of this new technology is bafflingly slow.

Businesses use Interactive Voice Response (IVR) systems because they save money – a lot of it - by directing callers to the (hopefully) correct department and freeing agents from having to handle so many calls by performing mundane tasks automatically, without human agent involvement.

Imagine how much more cash will be saved by implementing natural speech technologies! Not to mention the increased retention of high-quality, empathetic agents who generate rapport with those customers with whom they do deal, and the positive impression (a.k.a. loyalty, a.k.a. word of mouth marketing) from customers whose pet hates have suddenly been eradicated.

In the world of Contact Centres, natural language systems allow customers to serve themselves according to their own goals and directives. There is no need for customers to build up a complex mental model of their location within a service; they should be able to flit nimbly between declaring their goals and giving information.

What is more, with this technology, users can interact with an enterprise's business processes over various channels, such as the telephone, email or text message. They can even engage in an instant messenger session with a virtual agent, without necessarily even realizing there is no human customer service agent at the other end. What this means is that customers can use their own favoured gadgets to contact their suppliers – without having to queue to perform what is essentially a simple transaction, or being passed to a person for something off-the-wall or sensitive.

This last point is key. Given the choice, most people would still prefer to speak to a human or a human-like interface than a machine. There are two ways to address this issue. Virtual agents can be made to 'come to life': technologies are emerging which inject virtual agents with a sense of human ability, intellect, responsiveness and awareness – you can even animate them to give them a virtual 'presence' – you can give them give them anthropomorphic appearance. You can even give them subtle facial reactions to match the customer's vocal tone: a smile, a confused look, or concern, for example.

The natural extension of this humanising of virtual agents is to match up the human-like 'avatar' to the customer. Just as branches breed familiarity and – in most cases – trust in individuals, so this can be mimicked technologically. Virtual agents can have a personality (the one in our office is called Sarah, she is around her mid-twenties, has a West of Scotland accent, short brown hair and a pleasant, placid facial expression). The personalities are endless. A virtual agent (be it a vocal presence only, or a full-blown avatar) can be male or female, with almost any accent, any tone of voice, etc. Customers can create their own personalised virtual agent and (here is the clever part) can be matched to the customer when they return.

The other, complementary approach to a human's natural desire to deal with another human is to blend contact centre processes via workflow, such that self-service options interface seamlessly with human agents into a unified service capability.

The blending of human and virtual agents does not stop there. Virtual agents are ideal for handling the more mundane, repetitive, easily automated tasks. The phone call or text chat can then be handed over to a human agent who receives a transcript of the contact to date and can carry on the contact having been fully briefed (again, no need for repetition). Indeed, in the case of the text-chat channel, the customer need not know that they have been handed off from a virtual agent to a human agent, who may solve a problem before simply handing the contact back to the virtual agent.

This means that human CSRs should find themselves moving up the value chain, using their brain where the human brain is needed most, while having automated agents at hand to perform lower-level minutiae. A human agent is still required 'in the loop' for the rarer, and more complicated processes, such as conflict resolution, complaint and exception handling. I recently witnessed an agent spend twenty minutes calming down a distressed elderly customer; such wonderful customer service still requires the human(e) touch. A touch that Hal evidently lacked.