AI for people

Tapping virtual intelligence Using AI to make a dent in the skills crisis

The skills shortage is something that plagues almost every South African business, and on the surface, it shouldn’t be so. We have an abundance of people looking for work, but the issue of skills, and specific skills at that, is something that is holding back the country.

While there are some skill sets that require years of university education or technical specialisation, in many other cases, the training could be done on the job. Many companies are, however, reluctant to take on this burden because of the length of time it would take to get someone up to speed, and the risk that once they are au fait with the industries, they will simply be poached by competitors.

In the worlds of sales and customer service, the key remains getting new staff up to speed as quickly as possible. However, the range of options that many companies have today can make getting up to speed a lengthy process and sometimes new salespeople only have partial knowledge of the full product set.

Ryan Falkenberg, co-CEO of CLEVVA, believes there are ways to apply artificial intelligence (AI) to this problem, but in a way that puts people at the front of the solution rather than eliminating them completely. “The way many people think of AI restricts its scope to areas such as the interrogation of big data and machine learning, but we came at the area from a different angle. The issue is that big data can inform decisions, but you can’t necessarily control the conclusions that are reached. With those kinds of systems, the whole point is to find links in the data that may not be apparent to a human and trawl through data sets that are simply too large for a human to work through.

“What we find interesting is that a lot of the time, what we think of as expertise is actually the known logic that a person or team has built up over time. More often than not, companies are looking for people who are able to follow a known logic, “ he says.

Contextual knowledge

This knowledge may be very contextual, but it’s something that follows an established set of rules and it’s actually possible to capture these processes. The solution that CLEVVA has come up with falls under what it calls a Virtual Advisor (VA), which is, at its heart, a system for capturing the intelligence in a company and exposing it through a simple-to-use tool.

“If you’re working in a sales environment and you want to capture a virtual sales advisor, there is a formula, a fixed set of products with fixed features and benefits. At the same time, there is a fixed set of customer needs and the task of the VA is to bring to the fore the set of products that is most appropriate for the customer needs.”

Similar concepts do exist, but where CLEVVA stands out from the crowd is that the process of creating these systems has in the past relied on people hard-coding the business rules into the system. “The key to getting companies up and running quickly is to eliminate the need to code anything in order to create the VA,” says Falkenberg. “Also, how do you capture the decision process of an experienced employee?”

He gives the example of a field technician attempting to diagnose a problem with a specific piece of machinery. The key is that a technician with many years of experience will follow a logical process to diagnose a fault, examining the information at hand and eliminating possible causes until the solution presents itself. It is this kind of logic that the system strives to capture.

“We also understand that companies have different needs and differing levels of sophistication,” says Falkenberg. “One company may have a detailed understanding of its processes and simply want to digitise that process, while another may have been relying on the intuition of its sales team to understand what the needs of the customer are. So in one situation, we would be mapping a known set of processes, while in another, we may have to upload a massive product set with multiple variants and ask the system to decide what questions need to be asked in order to arrive at the correct outcome.”

Specific skills

The challenge is not access to information, as anyone who has tried to use Google to diagnose a technical problem will know, but rather, access to contextual expertise. Knowing what question to ask is much more valuable than having a set of manuals at your fingertips.

Falkenberg relates a story of one customer that had been servicing the commercial flooring industry, but was struggling with inconsistent levels of knowledge across its different sales channels because of the complexity of the product set and the demands of the customers. They required a high-level engagement, leaving little opportunity for inexperienced sales people to build up the knowledge base they needed. By capturing the sales logic inside CLEVVA, the company was able to get new salespeople up to speed in a matter of months rather than the two years it had previously taken. By creating a tablet-based sales tool, it was possible to allow the sales person to spend more time engaging with the client rather than simply trying to figure out want the correct option was.

In addition, by having a system that exposed the full product set, lesser known options saw greater uptake. At the same time, the system also allowed management to see a record of the engagement and monitor the performance of staff much more closely. The original thought behind the company was not to replace people with technology, but, rather, to augment human skills with technology that could perform specific skills better.

“SA has a skills crisis and the country is trying to train its way out of this by throwing huge amounts of information at people,” says Falkenberg.

“I got more interested in this and asked what we could do to fast-track the skills creation process. Part of the problem is that the way people are being taught in fundamentally inefficient. Simply throwing massive amounts of information at people and hoping they will assimilate it is counter-productive because there is a limit to how much information can be retained through traditional classroom learning. “The answer is that we are shouldn’t be trying to create more knowledgeable people; we should be trying to create more experts,” he says.

Job creation

By creating a VA that can harness exiting business logic as well as contain any information that might be needed, you eliminate the need for new employees to memorise every element of the company’s product set and rather focus on human interaction. Falkenberg says that experience has shown that using a VA does not detract from the human element of the business transaction; rather, it enhances this element, and clients tend to prefer the consistency they get from the system. “So now you can hire huge amounts of people without feeling that these new hires are a liability because they have to be trained up and are going to make mistakes. While some people see technology as something that reduces the potential for job creation, we believe that the VA is something that can stimulate job creation, enabling companies to bring in staff that would not have qualified previously, and, over time, build those people into experts themselves.” The issue of how a company creates these VAs is critical to the value that CLEVVA brings to the table.

Falkenberg says there is no requirement for coding when creating the virtual advisors. “Our aim was always to have something that people working in a business would be able to create. If you have to rely on coders to make a system work, then scaling the system becomes an infinitely costly exercise and deployment times stretch out.

CLEVVA offers a system that is not only easy to deploy, but cuts the amount of time needed to get a system up and running.”

Falkenberg relates the story of an HR team that brought 200 pages of flow diagrams to a meeting, with the diagrams describing the hiring processes for the organisation. Discarding the diagram, the team spent a few hours describing the hiring processes and after that, the CLEVVA team was able to deliver a working prototype.

Ryan Falkenberg and his brother Dane have been running businesses together for over 20 years. “After I left the consulting world in 1997, we started High Performance Learning, which was focussed on learning optimisation,” Falkenberg says. “The idea was to help people learn more in less time. At the time, we were one of the bigger privately owned learning consultancies in the country.

“We then started a company called Cuda Technologies because we were so frustrated by the inadequacies of learning management systems that were unable to perform when it came to the restriction the South African bandwidth placed on them. This meant that we could never get real-time learning; people always had to go to learning centres in order to do courses and this diminished the effectiveness of the process.

“As part of this, we realised that even though people were performing well in the learning environment and passing any tests that they were given within six months, their performance was nowhere near where we wanted it to be.

“The mistake was making the industrial-era assumption that data should reside in the human brain, but little of what people learn actually is retained in the long term. Also, the rate of change in the current world means that information that was learned just a few months ago changes and people keep having to relearn information, techniques and processes.”

The third problem was that the flow of business processes was becoming so complex that it was impossible for a person to retain all the knowledge they needed to keep track of the variety of options in any given situation. They realised that they were trying to get people to do things that machines were more adept at. They went looking for a technology that could realise their vision, a search that turned out to be fruitless.

After they decided to build their own, they came into contact with two brothers, Mark and Paul Pederson, who had a company called Base10. They were building CRM and ERP systems using powerful data architectures. In 2011, they bought the company and took the development team to form the basis of what CLEVVA is today.

 

© Copyright 2017. Brainstorm Magazine | Published by ITWeb Limited |

Click here to view the original article