The first time I talked to a chatbot, it didn’t go well. I had ordered a porcelain cup for a friend’s birthday. It arrived in 2 pieces because someone forgot the bubble wrap. I went to the website and the chat bot popped up immediately, inviting me to ask a question. So I logged my complaint. Unfortunately, the bot didn’t understand the phrase ‘the cup was broken’ and despite several tries to explain the problem, the bot didn’t offer me a solution. I finally called the support line, and an agent agreed to send me a (bubble wrapped) replacement in a matter of minutes.
Since that first experience, bots have come a long way, not in the least because natural language processing (NLP) and machine learning (ML) technologies have improved so much. Companies like Google, Apple, Microsoft and Amazon have introduced chatbots that people are willing to use at home. They are getting used to having bots around and talk to them “Hey Alexa, Hey Siri” if they need something.
And so, if you aren’t using HR bots yet, it’s time to seriously evaluate what’s out there. HR spends a lot of time handling employee questions. Often, these answers are readily available, and people can help themselves. Using bots to answer questions can alleviate the pressure on HR service teams and improve the employee experience. Bots are always on.
The future of service delivery
When you visit a website, often a little window pops up at the bottom, asking if you need help or have a question. That’s an example of a bot reaching out to you: “Hi, I’m Laura, how may I help?” Despite the name, there is no person involved. These bots automatically respond to something you do.
Chatbots can perform a variety of tasks, from serving customers in live chats to offering a download or selling a product via that same chat. And once they complete the task, they evaluate that interaction to have a better exchange with the next customer. They learn from their past to improve future conversations.
Chatbots range from very simple, providing ready-made answers to questions (“intents”), to increasingly complex, that use AI (artificial intelligence) to have intelligent conversations and build on that. And once they run out of options, there’s always the “warm handover” where a person steps in to finish the conversation.
While I recommend that you start with simple HR chatbots that provide answers to “how to” questions, you’ll want to evaluate the more intelligent bots too, because they provide a more comprehensive and better employee experience.
Never miss an article – subscribe to our monthly newsletter!
3 reasons to introduce virtual HR assistants
The first reason is obvious: chatbots save money by completing repetitive tasks without human interaction. They also save time: chatbots can handle these tasks much quicker than humans. They handle multiple customers simultaneously, and they don’t mind doing the same thing over and over again. Recii is an example of a company that helps recruiters identify suitable candidates much faster.
Secondly, chatbots provide consistency: an employee can ask the same question in many ways, using different words. “What’s my salary this month?” requires the same answer as “how much will I get on pay day?”. Chatbots look at the ‘intent’ behind the question and provide an answer based on what they’ve learned. We’ll look at Leena AI to explore automated employee interactions.
And then, there is the interaction itself: the conversation. Often, people are more comfortable talking to “non-humans”. They share things with machines that they would not discuss with a person. And so, understanding if people are happy at work and helping them stay engaged is what Max by Humaxa does best.
When I started talking to young companies that are developing solutions in this space, one of the first things I learned is that they don’t talk about chatbots: that’s just technology. They talk about virtual assistants because their solutions provide a service. Let’s take a closer look at how these virtual assistants are fundamentally changing HR service delivery.
Want a job? Interview with Recii first
Imagine you are a recruiter. You just posted a vacancy and you receive hundreds of applications to prescreen. Where do you start? You must vet these candidates by asking the same questions, just to qualify them for the next stage. It’s time-consuming: reach out, set up an appointment, have a conversation, and repeat that for every candidate. Some candidates will not call you back. And when you send an email, the response rate is low. If you don’t reach out, candidates will think you have “ghosted” them and never respond again.
What if you had a virtual assistant for the leg work? Recii was founded to do exactly that: automate the prescreening process to quickly help you narrow down the long list to a short list of qualified candidates. This process is relatively simple and repetitive, and an excellent use case for automation. It’s also scalable: the assistant can run interviews in parallel and handle dozens (even thousands) of people at the same time.
After a candidate applies, Recii responds with an email or text, and asks them to start the interview using their preferred channel of choice. The candidate can undertake it at a time convenient to them (even if that’s 3am) and use a medium they are comfortable with: chat, voice, whatsapp etc. They are also notified that this will be an automated conversation.
When candidates schedule the interview, that’s the first qualifier because they are willing to invest time. Candidates are made aware of the automated conversation and they don’t seem to be intimidated by it. Surprisingly, candidates are very transparent and helpful when answering questions, and often provide more information than they would in a personal conversation.
My initial assumption was that people would prefer text because no one talks on the phone anymore. The opposite is true: the majority of interactions is voice. Gareth Christian-Lim, CEO at Recii, observed that people are more comfortable to talk to machines (Siri, Alexa etc). They prefer speaking over typing because it’s quicker. And because people give more information, that’s helpful for the recruiter too.
Automated interviews are becoming more common, and not for blue collar jobs only: the solution has been used to fill a variety of roles from truck drivers and warehouse workers to developers and business analysts. A typical conversation starts with qualification questions that require a simple Yes/No. If a candidate doesn’t have the required certification, the interview stops and no one wastes time.
Based on the first responses, the assistant then moves on to the questions that require a more extensive answer. The recruiter only assesses candidates that get all the way through the interview. Recii’s clients use the solution so they can process large amounts of applicants with small recruitment teams.
While Recii initially focused on immediate hiring, some of their clients are taking a long term view and use it to nurture a pool of candidates. They can check in with these candidates once every 3 months to update their data (certifications, visa etc). In this way, a company can easily engage with candidates and build up their profile, so you know exactly when it’s time to reach out to them with a vacancy.
What about candidates who do not want to talk to a bot or get frustrated? While that happened a few years ago, it gradually changed. People are more comfortable talking to machines/devices and it’s quickly becoming the new norm. Gareth stressed the ethical side: candidates know they are talking to a virtual assistant. It’s all explained in the invite they receive to schedule the interview.
The trick is to ask relevant questions, so it’s clear to the candidate why they need to answer them. When the intent is simple and clear, the question receives good responses. People don’t seem to mind that they are talking to a machine. In fact, some people feel so comfortable, they share personal issues and elaborate in a way they wouldn’t in a first conversation with a recruiter.
Share your feelings with Max
To explore that topic a little further, let’s look at Max. Humaxa is using chatbots as a natural way to provide and receive feedback. We all use language, yet we find it hard to talk to our manager about problems at work and our professional development. If we make it less personal, would we share more?
Back in 2014, USC performed a study with a group of people suffering from PTSD. These people were reluctant to talk – they felt that having PTSD made them seem weak and incapable in the eyes of others. During the study they were asked to share their symptoms with a virtual human.
The “machine only” group opened up and volunteered uncomfortable and awkward topics freely. The group that filled in the regular questionnaire was far less forthcoming: they felt someone was judging them.
Other universities have repeated versions of this study in other fields. And the joint conclusion is: when things are awkward or uncomfortable, people are more willing to talk to a machine because they don’t feel judged.
As Carolyn Peer, CEO of Humaxa said: “It all comes down to trust: you don’t want to be perceived as weak or incapable. You don’t want to share intimate details with a person who might rate your performance and can influence your career.”
Humaxa provides Max. Think of Max as instant employee engagement. When you periodically run surveys, you’ll need time to process the answers, evaluate trends and address them. Max offers a live conversation with instantaneous evaluation and can give assistance or feedback to a person, with an actionable follow up.
If you run a diversity, equality and inclusion program, you can have Max reach out to employees and ask them what they think: do they feel included? Is the company doing a good job? And if the person responds positively, Max can immediately serve up a follow up question: “would you like to recognize someone who does a good job?” The data is analyzed anonymously, and insights are provided to the leadership so they can act on the findings. They can also use Max to dive deeper and further explore an issue that popped up.
Max is all about providing feedback. Max suggests actions based on what someone says, rather than run them through a funnel. It means that the assistant is constantly learning behind the scenes what works to improve the conversations. As shown in the example below, when an employee signals they are unhappy about something, Max can intervene and suggest a solution, in this case introduce the person to colleagues working on that topic.
As virtual assistants improve, they learn what works and move from answering questions to combine options and offer advice. When Max discovers that someone struggles with work/life balance, it can offer to find out how many vacation days that employee has, or if they need the name of a daycare provider for their kids. Finding out which action an employee takes, and if that helped, makes the assistant smarter and over time the advice becomes better and better. In that way, these assistants provide a more holistic HR solution.
Leena AI answers all your HR questions
The team behind Leena originally developed a platform to build chatbots without code so customers could create their own bots. But while observing what their customers actually built, Adit Jain, CEO of Leena AI, found out that these enterprises seemed very much focused on building internal HR bots to improve services delivery to employees. The team also realized that these projects were repetitive: the process was similar, and while there was some deviation in the content, there was a lot of overlap as well.
And so they went back to the drawing board and used the platform to build a generic, virtual HR assistant that can be used by all companies. The idea was that automation saves time by answering simple questions in a consistent way. As employee, you only click on your mic, ask a question and Leena provides you with the answer.
As Leena grew, so did its knowledge base. And the team discovered that there were more applications than they originally thought. Today, employees can ask Leena all kinds of questions pertaining to HR services, and Leena will answer them based on the policy and procedures it knows, personalized to those of the company.
Leena started with a basic HR knowledge base, and over time the assistant learned that clients have slightly different policies and so understands nuances. When new clients sign up, Leena ‘reads’ their policies, compares them to the existing knowledge base, and then serves up questions and answers that HR professionals only have to review. Leena also connects to the existing HRIS to answer questions about employee’s personal situations like vacation balances, salary payments and so on.
Leena now serves over 100 customers, helping more than 500.000 employees in 15 languages. One of those clients ran a service center with 100 agents. As Leena became more knowledgeable, the company was able to reduce the staff to 65 people.
As part of the continuous learning effort, whenever the assistant receives a question or request that is unique, meaning, Leena hasn’t seen it before and can’t use the knowledge base to derive the answer, it automatically creates a ticket in the customer’s case management system and routes that to an HR professional for evaluation. The response is then added to the knowledge base, so Leena learns something new and can answer the question in future.
Leena AI is one of the companies that benefits from the pandemic. HR professionals have a need to manage services better while employees work remotely, but companies are nervous about hiring more staff. Intelligent assistants like Leena can help them provide these services without increasing the number of service agents.
With all that HR is being asked to do now, it’s imperative they use automation if they want to continue to deliver quality services. Companies simply don’t want to hire more people, but they do want to acquire technology to automate services, as that is seen as a future-proof investment.
Let’s talk about ethics…
In 2018, Google went public with their latest advances in voice assistants: it called to a salon to make an appointment. At first everyone was wowed, but pretty soon the question became: is it ethical to have a human interact with a machine without telling them?
AI can be useful, but people find aspects of the technology scary and intimidating. Comparable to my first chatbot interaction, people have bad experiences. All 3 founders stressed that applications can’t be creepy. As employer, and especially as HR, you have to be transparent about your plans and applications, and disclose what these assistants can and can’t do. Just as recii makes it immediately clear to a candidate that they will be interviewed by an automated system.
Employees sometimes ask if the assistant reads (meaning: understands) all conversations. It’s important to explain that automated systems don’t understand it like a human would – they simply processes the interactions and learn how to respond. There’s no deeper meaning involved. Sometimes customers can also be apprehensive: “Is your assistant going to advice my employees to find another job?”
It’s also about behavior: some people yell at bots. While that sounds bad and people shouldn’t do it, the assistant has no feelings and doesn’t care. Carolyn Peer brought up a surprising finding: people actually sympathize with their virtual assistant. When they answer questions, they use its name. And they do that because they want to be polite: ”It’s rude to talk to them and not use their name”.
… and not forget bias
A conversation on chatbots isn’t complete without mentioning prejudice and unconscious bias in AI. Remember Tay, Microsoft’s infamous attempt to launch an AI chatbot on Twitter? Tay was supposed to learn from her interaction with humans. Unfortunately, those humans decided to turn Tay into a pro-Nazi, racist and anti-feminist bot, and Microsoft removed her within 24 hours.
A chatbot is trained with learning algorithms and neural networks and will incorporate the prejudice of its builders. It’s imperative that you understand the design principles behind the solution. The vendor must be transparent about the algorithms, and you must keep track of what happens after it goes live. Just as you would monitor a starting human assistant, you’ll need to know what a virtual assistant is up to as well.
Imagine that the first 5 employees that have a work/life balance question all stop the interaction once the assistant suggests childcare options. As a coincidence, these first employees were all female, and they were all satisfied with the answer. A virtual assistant could conclude that it should always suggest childcare as first option to women with a work/life balance question. From then on, It would give gender biased responses, without you being aware of it. It might seem an obvious example (and it is), but it’s critical to periodically review the questions and answers to ensure that your virtual assistants are on the right track.
Are bots the future of HR?
Gartner predicts that we should consider AI-driven chatbots as “the future of everything that’s happening in HR.” We’re not there yet, but recent advances in technology are getting us close very rapidly. And people seem to be accepting bots as a normal way of interacting with services organizations. The next step is zero UI, where employees will ask questions using voice instead of devices. Conversational technologies will play a big role.
I would recommend that you operate under the assumption that your future HR service delivery will be driven by bots. There’s no good argument that it wouldn’t and it has a number of benefits that you cant achieve another way. With that in mind it’s time to run your own pilot. As the examples above show, there are virtual assistants available for every aspect of HR.
There’s another important reason why you should look at conversational platforms: the 2020 pandemic has increased the adoption of digital work environments, like Teams, Slack and other productivity platforms that I wrote about a year ago.
Vendors are integrating their services within these workplaces, using API’s, so employees can seamlessly use backoffice services like HR or IT without switching to other solutions. Employees ask their question in the workspace, and the backoffice will ensure the request or transaction is processed by the corresponding service.
So it’s unavoidable that there will be bots in your future. High time to get ready…
Never miss an article – subscribe to our monthly newsletter!