Chat Bot HR ARAGO Consulting

Are Savvy HR Chatbots able to fulfill the expectations?

Previously in our blogpost “Powerful HR chatbots offer help!”, we spoke about the exciting and continuous journey of the digital world. Often, thinking of solutions for this challenge could be easier than anticipated. Alternative solutions for digitalization could be easier than anticipated. We mentioned HR chatbots and conversational artificial intelligence (AI) as an interesting path to consider. We want our employees to have tools at their disposal that speak for themselves and require little to no upfront knowledge. They should be fast and to the point in terms of what you want to achieve with them. The application should be easy to deploy on multiple devices in order to accelerate adoption. Program sponsors will likely throw in that whatever solution should be cost optimal to develop, scalable to roll out and easy to maintain. Whether HR chatbots are able to fulfill all those expectations is obviously not a black or white question, let alone the answer.


Hey Siri, thanks for boosting the user experience!

Chatbots are not completely new. Recent years have shown an increase in the number of enterprise websites that have a chatbot available to support different scenarios. Although the technology is in essence very similar, they serve a lot of different purposes. We can buy a wide variety of products through them (ranging from food and clothes to train and flight tickets), search assistance in selecting the right product, ask questions about regulations, check the weather forecast. A big boost came with the development of next generation conversational AI such as Apple’s Siri or Amazon’s Alexa Skills platform.


Unfortunately, a lot of publicly available bots have been badly setup and detract usage after a couple of interactions. In most cases, that is related to an unclear vision about what companies want to achieve with a bot and how they can keep it growing and staying up to date with change. At the same time, we notice the level of maturity of a bot pre go live is seldom high enough.


Quickly familiar with HR Chatbots

Though setting a vision about conversational applications might seem complex at first sight, it can easily be stripped of a lot of confusing side-tracks by putting the customer experience at the centre of each deliberation. In the context of HR related use-cases, the same is valid. Think of your employee, not just once at the beginning of a project, but continuously and carefully. Before entering the difficult discussion about viability of a business case, let’s try to understand the added value from an employee perspective. The true calling of conversational user experience is in allowing an easy and always on access to both transactions and knowledge within an enterprise. The way of facilitating that via an HR chatbot could be through clickable dialogues, words, speech or even a mix of all of the aforementioned. The big advantage is that most employees are familiar with peer-to-peer chat platforms. They are all around us and therefore present hardly any hurdle to overcome. With the generation of digital natives starting to become part of your organizations, messaging will be even more intuitive and natural.



HR is in a position to perfectly meet this new paradigm. As a user interface, it is hardly impossible to think of a use-case that couldn’t be achieved via conversational AI – provided it is linked to a modern (read API enabled) backend system. Some examples of area’s that are already in use in multiple enterprises:


1) HR FAQ’s allowing employees to ask their question to a virtual assistant rather than relying on a more static search. The HR chatbot can take the employees through its dialogue skills by the hand and offer additional information to a related topic or kick of a process linked to the query;


2) Updating personal information from one single user interface, supported by integrations to multiple downstream systems in order to skip asking for any redundant and already known information;


3) Requesting time off as an employee and approving as a manager, at any time from any device. The Digital HR Assistant can take care of subsequent activities such as informing co-workers, setting up out of office assistants, alerting for time off that is soon to expire;


4) Expense Claim submission on the spot: subscriptions to learning courses, alert for mandatory trainings, notifying of new trainings.


Consolidating the existing HR applications

The beauty of a conversational interface is in fact the combination of multiple use cases into one single medium. That is truly a golden bullet in terms of user experience but will also allow to bring a next step in the consolidation of existing applications. Scalability and cost efficiency will eventually follow. When truly pursuing a vision of working with conversational AI, it should be considered as a program rather than a single project. Taking the assistant gradually to a next level but starting with a solid foundation is the only way forward. But you cannot let it rest if you want to avoid it becoming just a gimmick.


Without having the ambition to detail completely how to start building an HR chatbot, following activities will be part of the foundation setting:


1) Determine the channels you want to build or leverage to allow employee “access” to the HR bot and integrate them with the chat platform;


2) Identify the intents (chatbot lingo for “instructions”) you will want your HR chatbot to understand and decide on any additional business logic you require (dialogues or deeper logic);


3) Feed those intents with potential user expressions and start training the bot by fine-tuning the language model;


4) Evaluate the downstream integrations needed to support the use cases you have in mind and re-use existing interfaces to reduce time to market.


Although some of those steps are rather conventional in nature as they rely on analysis and development, the pitfall might be in steps 2 and 3. Most of the IT we have been building doesn’t require a natural language model. But see it as a necessary step – IT can adapt to humans, instead of the traditional other way around. How we can make this step happen will be elaborated in our next blog.