85%

OF ALL CUSTOMER RELATIONS WILL BE AUTOMATED BY 2020

‘Chatbot’ is one of today’s buzz words. These digital conversational agents are now being touted as the future of customer relations. Already robots direct, advise, and guide Internet users 24/7 and this functionality works in tandem with traditional customer contact channels such as call centres, email and live chat. Nowadays people use chatbots to book a train ticket, order in a pizza, make appointments and so on. According to US research and advisory firm Gartner, by 2020 85% of all dealings with customers will no longer require any human interaction. However, virtual agents are not entirely new. The first bot goes back all the way to 1966. The Eliza programme developed by MIT Professor Joseph Weizenbaum was confined to reformulating suggestions from its interlocutor into question form, but at the time even this limited functionality caused quite a stir.

Since then, artificial intelligence has progressed by leaps and bounds and the Internet giants have given their audiences a new taste for this type of functionality as they vie with each other to bring out new tools. Facebook took a giant step over a year ago in providing developers with a Software Development Kit (SDK) for Messenger, its instant messaging system. The result: 33,000 bots created within just a few months. And having won over consumers, chatbots are now being redirected towards companies’ internal clients. They are now able to successfully fulfil a number of low-value-added tasks for support departments including the help desk, the general services department and the Human Resources department.

HR chatbots aimed at banks, energy sector companies and public sector organisations

The Human Resources department is certainly a prime area for such tools, given the potential for, and advantages of, automating all those frequently asked questions from the staff. How many days of leave do I have left this year? How many overtime compensatory days am I owed and can I add those days to my vacation total? How do I fill in my expense report? These are the sort of questions that HR people have to field day-in, day-out. Benoit Capitant, Managing Partner i/c Information and Communication Technologies for Human Resources programmes at Paris-based Information Systems consultancy mc2i Groupe, says that “HR chatbots are designed for companies with more than 10,000 employees which have a high flow of questions and requests – e.g. in banking, the energy sector and the public sector.” In France, Axa, Groupama, Société Générale and PSA have already taken this step.

CHATBOTS IN HR

Chatbot

Shutterstock

‘Self-service’ HR

The way it works is that instead of picking up the phone, filling out a form or sending an email to the Human Resources department, a company employee can put a question to the chatbot using natural language. The bot can be contacted through the instant messaging service used by the company, such as Skype, Slack, Google Hangouts or Messenger. Once the employee logs on to the in-house system, s/he will be automatically identified from the company personnel directory. “Integration of the chatbot with the IT system may be highly advanced or less so, depending on the sophistication of the service required,” points out Pierre Lanctuit, a consultant at mc2i Groupe, explaining: “A request for leave will follow the traditional workflow system. It will be forwarded to the manager, who will decide whether to approve it or not.” 

On the other hand, for a Wiki-type chatbot, integration into the IT system may be negligible. With this approach, an employee can put questions regarding ‘dry’ or ‘static’ information such as the tasks of the Works Council, details of the latest Collective Labour Agreement, or the procedure for submitting an expenses claim. Answers are provided by searching a static knowledge base. Explains Cyril Texier, co-founder and editor of Do You Dream Up: “As a general rule, the information requested by the employee is already published somewhere on the company intranet, and the chatbot just makes it easier to find. Basically, you’re just switching from pull mode to push mode.” 

 The degree of intelligence possessed by the chatbot is based on natural language processing, a field which is situated at the crossroads between IT, artificial intelligence and linguistics. Basically, this means the machine’s ability to grasp what is behind a question or request by picking out the keywords and analysing the context in which the sentence is formulated. Once the questioner’s basic intention is known, the bot will either search for an answer in a database or launch an action.

SMART CHATBOTS TO SELECT JOB CANDIDATES

Chatbot RH

Shutterstock

The advantage for the HR department is that they are operating within a closed linguistic field where 80% of all questions asked are the same,” underlines Cyril Texier, explaining: “Take for example a defective entry badge. The relevant questions are highly predictable. Have you lost the badge? Is it broken? And you know in any case that this type of badge has to be changed on average every two years.” Any unrecognised questions are forwarded to the relevant business or department. They’ll have to decide whether the question can be linked to a similar question that has already been handled before or whether it’s necessary to create a new database item. 

And when it comes to chatbots, the experts do not see any generation gap. The ‘digital natives’ who are thoroughly used to having chat sessions on WhatsApp or Telegram and have lost the habit of picking up the phone will immediately be in their element. However, their older colleagues also tend to see the convenience value in using this kind of instant conversational interface.

Initial filtering of job candidates

Quite apart from this ‘self-service HR’ approach, chatbots are now being called upon to carry out a range of other tasks. Salary matters are high on the agenda – e.g. When will my next salary payment be on my bank account? May I have an advance? – as is the whole field of recruitment. By using questionnaires and running a series of tests designed to verify that job applicants really do possess the skills claimed on their CV and share the company’s core values, the bot is able to carry out the initial candidate filtering process. One example is the bot used by French startup Pipplet to test job applicants’ English skills. The bot can even go a step further by conducting a virtual pre-interview.

Pipplet
  • 1 min

 Says Pierre Lanctuit: “If the candidate comes through the first stage successfully, the robot can invite him/her to answer a few questions in front of the webcam and send the video for the recruiter to watch later.“ The bot can also send a personal reply to each of the unsuccessful applicants. And while French companies may still be hesitating to take this step, many of them are now thinking seriously about it. Do You Dream Up has received a lot of requests from CAC 40 (major listed companies in France) on this subject.

Real-time follow-up of candidates

Belgium-based global audit, accounting and consulting group Mazars is now using a bot called SAM, which was developed by the FCINQ agency, as a first step towards recruitment of new staff. This Facebook Messenger-based coach offers advice to potential job applicants in such areas as preparing for an interview. Mazars, which hires around 1,000 new staff every year, mainly young graduates of universities, business schools and engineering colleges – who are much sought after in the professional world – sees SAM as a way of helping to boost its employer brand in the face of the current war to recruit talent. “We wanted the tone of this chatbot to reflect our corporate culture – a fun, laid-back atmosphere,” stresses Olivia de Faÿ, Head of Talent Acquisition and Employer Branding at Mazars. And the bot has certainly proved popular. Launched in April, SAM generated over 3,000 conversations in just five weeks, resulting in some 1,000 re-directions to job vacancies at the Group. Even better, Mazar’s recruitment target group are showing up. Statistics provided by Facebook show that the 18-25 age-group is the most represented, followed by 26-34-year-olds.

Chatbot Mazars
  • 1 min

Chatbots pumped up on AI will soon be able to find you that dream job!

So what’s the next step? Cyril Texier reckons that advances in artificial intelligence, in which “we’re only just at the very beginning”, are opening up a wide field of opportunities. “Depending on an employee’s profile and career development hopes, a chatbot will be able to recommend a training programme or offer another job at the company,” he foresees. 

Nevertheless, our experts underline that companies should enable chatbots to play to their natural strengths – providing an immediate, appropriate response 24/7 – but without any false pretences: a chatbot should never try to pass itself off as a real person. And if the bot cannot find the answer, it should quickly hand over to a flesh-and-blood HR assistant. On no account should there be any attempt to de-humanise human relations. A bot is not there to replace human recruiters but to assist them. The Man-Machine duo should be working as a team here, rather than one looking to replace the other. HR managers and other recruitment specialists are not about to disappear. It’s just that their jobs are going to change and evolve in such a way that they will be able to devote more time to the central aspect of their profession: human relations.

By Xavier Biseul